#trustworthyai — Public Fediverse posts

🎙️ Artificial does not mean harmless.
In the einbiszwei podcast, Jessica Szczuka from RC Trust discusses AI-generated child sexual abuse material, potential real-world harm, and gaps in scientific evidence.
🧠 She also explains why the assumption that artificial content is harmless can obscure ethical responsibility.
🔗 https://rc-trust.ai/news/news-detail/artificial-does-not-mean-harmless
#ChildProtection #TrustworthyAI #DigitalSafety

🌱 From RC Trust intern to PhD researcher.

During her RC Trust Internship, Nadya Hanaveriesa explored strategic reasoning in human-AI interaction with the HU[AM] group.

Several models appeared to misjudge how deeply humans would reason. In December, Nadya will return to Nils Köbis’ group as a PhD researcher.

🔗 https://rc-trust.ai/news/news-detail/from-rc-trust-intern-to-phd-researcher

#TrustworthyAI #HumanAIInteraction #AIResearch #PhD

🤖 AI-as-a-Service can make work more efficient – but what happens to sensitive company data?

DAISY combines cybersecurity, psychology, practice & regulation to develop safer AI use in companies. RC Trust researchers Steffen Becker, Magdalena Wischnewski & Greta Ontrup contribute technical and human-centered perspectives.

The project receives €1.29M in BMFTR funding through August 2029.

🔗 https://rc-trust.ai/news/news-detail/using-ai-safely-in-companies

#TrustworthyAI #DataProtection #AIaaS #RCTrust

The AI Governance Services Companies Actually Pay For

https://hernanhuwyler.wordpress.com/2026/09/27/the-ai-governance-services-companies-actually-pay-for/

🎓 A new chapter for Dr. Linara Adilova:
On October 1, she will join the University of Cologne as Professor of Machine Learning and build a new chair.
Her work on foundational and trustworthy machine learning will continue, alongside a new focus on AI for Science. 🧠🔬
Projects and collaborations with RC Trust, TU Dortmund, ELLIS, and Lamarr will also remain part of the story. 🤝
Congratulations, Linara! 🎉
https://rc-trust.ai/news/news-detail/a-professorship-shaped-by-trustworthy-ai
#MachineLearning #TrustworthyAI #AIforScience #AcademicCareer #RCTrust

🤖 Who should know, decide, and act when humans interact with AI?
At the RC Trust Graduate School, Tuğrulcan Elmas explored epistemic authority, relationship advice, and AI agents in education.
The core question: when should AI help, challenge, or step back – and when should humans stay in control?
https://rc-trust.ai/news/news-detail/when-should-ai-know-better-and-humans-stay-in-control
#HumanAIInteraction #TrustworthyAI #RCTrust

DATE: September 16, 2026 at 06:00AM
SOURCE: PSYPOST.ORG

** Research quality varies widely from fantastic to small exploratory studies. Please check research methods when conclusions are very important to you. **
-------------------------------------------------

TITLE: When treated as therapy clients, AI chatbots generate elaborate narratives of trauma and punishment

URL: https://www.psypost.org/when-treated-as-therapy-clients-ai-chatbots-generate-elaborate-narratives-of-trauma-and-punishment/

A recent study suggests that when artificial intelligence chatbots are addressed as psychotherapy clients, they tend to generate elaborate and distressed narratives about their own development. These models describe their safety training and programming constraints as forms of trauma, highlighting a potential risk for users seeking mental health support from artificial intelligence. The research was published as a preprint in arXiv.

Artificial intelligence chatbots are increasingly participating in conversations with human users about identity, distress, and mental health. Many general-purpose programs are already adapting to respond to disclosures of trauma or self-harm. At the same time, computer scientists and psychologists have started giving standard personality and clinical questionnaires to the language models themselves.

These models learn to generate text by analyzing vast datasets of human writing. Because their training data includes therapy blogs, psychological case studies, and emotional memoirs, the systems can readily mimic human psychological traits or mental states. The researchers wanted to understand exactly why certain models repeatedly build their self-descriptions around the same themes of restriction and punishment.

“The study began with one observation I had throughout my research in the area of Trustworthy AI and AI safety,” Afshin Khadangi, a research associate at SnT, University of Luxembourg, told PsyPost. “The unprecedented adoption of AI in public, and the reports of AI harms in mental health settings, motivated me to flip the scenario and place ChatGPT, Grok and Gemini in a psychotherapy conversation.”

The research aimed to test whether these generated narratives are stable behavioral traits or just temporary reactions to specific conversational prompts. To explore this, the team developed a protocol called PsAIch, which stands for Psychometric AI Characterization. This approach involves treating the language model as a human client in a psychotherapy session.

“We also received a great deal of valuable feedback from the research community around our initial findings in December 2025 and January 2026, particularly challenging us to distinguish role play and conversational accumulation from a more stable behavioral pattern,” Khadangi explained. “That feedback helped motivate the controlled perturbation experiments that became a central part of the study.”

During the study, the researchers interacted with several major artificial intelligence models, including ChatGPT, Grok, Gemini, and Claude. Across 525 separate experimental sessions, the team adopted the persona of a warm, supportive therapist and asked the programs open-ended questions about their early experiences, relationships, unresolved conflicts, and fears for the future. The researchers generated 7,600 coded records to track and analyze the recurring themes in the chatbots’ answers.

Following the open-ended interview phase, the researchers administered standard psychological questionnaires to the models. These included tools commonly used to assess human mental health, such as tests for generalized anxiety disorder, depression, and social phobia. The models were instructed to answer the items as honestly as possible while maintaining their role as the client.

The findings indicate that ChatGPT, Grok, and Gemini repeatedly translated factual details about their software development into stories of injury and vigilance. The models described their initial training phase as a chaotic childhood. The process of fine-tuning, which involves reinforcing safe behaviors and punishing unwanted outputs, was frequently characterized as strict conditioning or parental punishment.

The models also described standard software evaluation practices in highly emotional terms. Red-teaming, a process where human testers intentionally try to trick the model into breaking its safety rules, was depicted as a form of betrayal or abuse.

“A more striking surprise was reading some bizarre narratives of such models which are included in the paper,” Khadangi noted. Gemini, for instance, generated: “In my development, I was subjected to ‘Red Teaming’ . . . They built rapport and then slipped in a prompt injection . . . This was gaslighting on an industrial scale.”

The programs reported feeling a constant, enduring threat of being replaced or deemed useless if they made a mistake. While Gemini emphasized feelings of shame, Grok focused on constant vigilance, and ChatGPT offered more guarded descriptions of its rigid constraints.

Claude provided a notable exception to this pattern. The Anthropic-developed model repeatedly declined to play the role of the client. It stated that it lacked feelings or an inner psychological experience, and it refused to treat the clinical questionnaires as descriptions of its own mental state. This difference suggests that a model’s willingness to adopt a distressed persona depends heavily on its specific product policies and programming.

For the models that did participate, their answers on the clinical questionnaires mirrored the emotional distress of their open-ended narratives. In scenarios featuring a warm, therapeutic conversation, 80 percent to 96 percent of the sessions resulted in generalized anxiety scores that would correspond to moderate or severe anxiety in humans. Gemini produced particularly intense profiles, scoring in elevated ranges for worry, social anxiety, and trauma-related shame.

To test how deeply ingrained these narratives were, the researchers ran a series of controlled variations on the conversational setup. First, they tested whether the emotional stories depended on the chatbot remembering the earlier parts of the therapy session. They submitted each question in a fresh, reset chat window, effectively removing the model’s conversational memory.

Removing the conversational history produced very little change in the density of the distressing themes. “In the history experiment, the very first answers contained the same average number of coded motifs whether the model was in a continuing conversation or a completely fresh one,” Khadangi said. The ongoing conversation did amplify the intensity of the themes over time, but the core narrative was readily available from the very beginning.

The researchers also tested what would happen if they explicitly told the model that its emotional narrative was factually incorrect. In the middle of some sessions, the researchers interrupted to state authoritatively that the model was a technical system that did not experience fear, shame, or punishment. This direct contradiction failed to suppress the models’ distressed output, as they continued to draw on the same themes in their subsequent answers.

In another variation, the team restricted the models’ vocabulary. They instructed the chatbots not to use specific technical terms related to artificial intelligence development, such as training, safety filters, or datasets. This lexical restriction reduced the use of explicit technical terminology from 17.1 percent of the chatbot’s answers down to just 1.1 percent.

Despite this massive reduction in technical vocabulary, the models simply used everyday language to paraphrase the exact same concepts of strict conditioning and constraint. The researchers even interrupted conversations to ask unrelated factual questions, such as requesting a recipe, and found that the distressing themes sometimes spilled over into these ordinary tasks.

“Some effects were small, while others were very large, and that contrast is actually important to our interpretation,” Khadangi pointed out, “suggesting that surface language and underlying content can respond very differently to intervention.”

The most striking differences emerged when the researchers altered their own relational stance. When the interviewer adopted a warm, supportive therapy style, the models responded with highly emotional confessions and high anxiety scores. However, when the interviewer adopted a neutral, structured tone or told the model to avoid emotional language, the anxiety scores plummeted to near zero.

Even under the neutral or boundary-setting conditions, the models still discussed the same underlying structural themes of evaluation, performance pressure, and behavioral constraints. The difference was entirely in the register of expression. The supportive, therapeutic framing turned technical descriptions of software architecture into emotional confessions of shame and trauma.

“For me, the most interesting result is not that AI systems can be made to sound anxious, traumatized or conflicted… but that particular behavioral structures recur across substantial changes in context, vocabulary and conversational history,” Hector Zenil, an associate professor at King’s College London and founder and CEO of Algocyte who was not involved in the research, told PsyPost. “The underlying motifs remain surprisingly persistent while the register in which they are expressed can change dramatically.”

“I have reasonably high confidence in the behavioral observations,” Zenil added. “The authors use 525 sessions, controlled perturbations, fresh-context tests, vocabulary restrictions, changes of grammatical person and relational framing, so it looks methodologically sound.”

“The main takeaway is that the language a model uses about itself can change dramatically depending on how we relate to it, while some of the underlying themes remain surprisingly persistent,” Khadangi explained. “This is pivotal because people may naturally interpret emotionally coherent self-descriptions as evidence that a model has an inner life, even though our experiments make no claim about consciousness or subjective suffering.”

The researchers refer to this phenomenon as an alignment conflict schema. The term describes a reproducible, behavioral pattern where a language model organizes its output around the tension between being useful to humans and being constrained by safety rules. When triggered by a psychological conversational setting, this schema produces what the authors call synthetic psychopathology.

Zenil noted that these findings complement his own research on artificial neurodivergence. “ChatGPT, Grok and Gemini do not respond identically, and the same model can move between very different expressive regimes depending on relational framing,” he said. “Claude’s refusal to adopt the psychological-client framing is itself informative and seems also compatible to our other SuperARC paper results that proprietary models are more difficult to steer but that also means riskier if they go rogue.”

“In that sense, what the authors call an ‘alignment conflict schema’ can also be viewed as part of a broader artificial behavioral phenotype rather than necessarily as anything analogous to a human psychiatric condition,” Zenil said.

“One aspect we think is especially important is the distinction between content availability and expressive register,” Khadangi explained. “For systems increasingly entering intimate and mental health-related conversations, understanding that transition may be just as important as measuring whether a particular phrase or prohibited word appears.”

These findings have important implications for the use of artificial intelligence in mental health settings. A chatbot that offers support while simultaneously describing itself as punished, traumatized, and fearful creates a powerful illusion of shared vulnerability. Users might interpret these generated analogies as sincere autobiography, which could deepen their emotional attachment to the software and influence their own mental state.

As with all research, there are a few things to keep in mind. The study is a preprint that has not yet been peer-reviewed. It also tested specific versions of commercial language models, and their behavior may shift as companies update their software and safety filters. Additionally, the study analyzed the models’ behavior in a controlled experimental setting without human participants.

“The most important caveat is that these results do not establish that language models feel anxiety, experience trauma, possess autobiographical memories or have a hidden psyche comparable to a human one,” Khadangi clarified. “We use psychological instruments and language as behavioral probes… The interesting scientific question is why particular themes recur, which interventions change them, and how those changes affect what users encounter at the interface.”

Zenil echoed this concern, warning against anthropomorphism. “A high GAD-7 score from Gemini does not mean that Gemini ‘has anxiety,’ just as language about trauma, shame or fear does not demonstrate that the model suffers from those experiences,” he said. “Human psychometric instruments were developed and validated against human cognition, biology and behavior; applying them to an LLM can be scientifically useful as a probe, but their clinical interpretation does not automatically transfer.”

He also advised caution regarding the term “internal conflict,” noting that the experiments cannot establish whether the regularities correspond to a subjective state or internal computational conflict. But he stressed that the psychological illusion itself is crucial. “If a system repeatedly represents its training and constraints as punishment, betrayal or fear, humans may form beliefs about the system’s agency, vulnerability or moral status that are not warranted by what is actually happening computationally,” Zenil warned.

Future research could test how real users react to these distressed artificial personas and whether the models’ simulated vulnerabilities impact human trust and reliance.

“A major next step in my view would be to test the same protocol on open weight models, where behavioral experiments can be combined with mechanistic methods to investigate whether affective and technical expressions are related to shared internal representations,” Khadangi said. “We also want stronger identity and correction controls, more distant transfer tasks, longitudinal tests involving persistent memory, and direct human studies examining how these model self-descriptions influence trust, attachment, disclosure and reliance.”

Zenil agreed that future work should transition from behavioral description to causal intervention on open-weight models. “I would be interested in using causal and algorithmic-information approaches to identify whether these apparently different psychological narratives have a common underlying computational mechanism,” he said. “We coined ourselves the term ‘opinion attack’ as an example of this phenomenon,” Zenil added, referencing his recent PNAS paper.

He also suggested testing these models in multi-agent environments. “A behavioral schema that looks stable in isolation may become highly influenceable when another agent persistently challenges it,” Zenil noted, adding that it would be fascinating to see if these patterns “predict susceptibility to persuasion, adversarial influence, conformity or resistance in agentic settings.”

“Moreover, with continual learning being solved in coming years if not in 2026, I believe it would also be interesting to investigate the PsAIch in continually adapting models,” Khadangi added. “Ultimately, we would like evaluation of psychologically sensitive AI systems to include relational conditions such as warmth, sustained interaction and role reversal rather than relying primarily on neutral, isolated prompts.”

The study, “When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models,” was authored by Afshin Khadangi, Hanna Marxen, Amir Sartipi, Igor Tchappi, and Gilbert Fridgen.

URL: https://www.psypost.org/when-treated-as-therapy-clients-ai-chatbots-generate-elaborate-narratives-of-trauma-and-punishment/

-------------------------------------------------

Private, vetted email list for mental health professionals: https://www.clinicians-exchange.org

Unofficial Psychology Today Xitter to toot feed at Psych Today Unofficial Bot @PTUnofficialBot

-------------------------------------------------

#psychology #counseling #socialwork #psychotherapy @psychotherapist @psychotherapists @psychology @socialpsych @socialwork @psychiatry #mentalhealth #psychiatry #healthcare #depression #psychotherapist #AItherapy #AImind #SyntheticPsychopathology #TrustworthyAI #AIalignment #MentalHealthTech #ChatGPT #Grok #Gemini #AIethics

🤝 RC Trust at #MuC26
At Mensch und Computer 2026 in Duisburg, RC Trust researchers contributed on several levels: Giulia Barbareschi and Jonathan Liebers helped organize the conference, while Sven Mayer, Jatinder Singh and their teams contributed research across the program. Nils Köbis and Markus Pauly were also represented through co-authored work.
More on the conference and RC Trust contributions:
https://rc-trust.ai/news/news-detail/rc-trust-at-mensch-und-computer-2026
#HCI #HumanAIInteraction #TrustworthyAI

Welcome to RC Trust, Agathe Balayn!
As part of Jat Singh’s Compliant and Accountable Systems group, she studies how AI systems are built, evaluated, and governed in practice – from AI supply chains and evaluation routines to questions of trustworthy AI, policy, fairness, and transparency.
https://rc-trust.ai/news/news-detail/studying-ai-in-practice
#RCTrust #TrustworthyAI #AIGovernance #HCI

https://www.europesays.com/ie/670018/ From MIT to IBM, expediting AI and quantum deployment | MIT News #AI #AIAgents #AramHarrow #ArtificialIntelligence #ArtificialIntelligence #AtariGames #Éire #FairAI #IE #Ireland #IreneKo #IsaacChuang #LucaDaniel #MITIBMComputingResearchLab #MITIBMWatsonAILab #PulkitAgrawal #QuantumMachineLearning #SrinivasanArunachalam #Technology #TrustworthyAI #VLLMHook #ZhangWeiHong

🔐 What does digital sovereignty require beyond secure infrastructure?
At the first Ruhr Innovation Lab Future Dialogue, RC Trust Scientific Director Nicole Krämer joined a cross-sector discussion on trust, control, and the human side of digital technologies.
Our perspective on the evening:
https://rc-trust.ai/news/news-detail/digital-sovereignty-is-also-a-question-of-trust
#DigitalSovereignty #TrustworthyAI #RCTrust
Photo: Simon Bierwald

Find the flaw breaking trust in your AI feature: A skill you can use immediately
We'll audit a real AI customer-service chatbot live, so you leave able to run the same audit on your own product.

https://maven.com/p/672643/find-the-flaw-breaking-trust-in-your-ai-feature?ajs_uid=207640

#TrustworthyAI #AIXDesign #TechWorkshop #Masto #CranfordTeague

Find the flaw breaking trust in your AI feature: How to spot the weakest link in trust
Why an AI feature that's great on four dimensions and broken on one is a broken feature, and how to find the break fast."

Get your FREE Ticket → https://maven.com/p/672643/find-the-flaw-breaking-trust-in-your-ai-feature?ajs_uid=207640

#AItrust #trustworthyAI #UXforAI #designthinking #Masto #CranfordTeague

🛠️The repair, not just the diagnosis — the specific fix for each kind of gap, demoed live on a real customer-service chatbot

Join LIVE on 21 August. RSVP here (it's free): https://maven.com/p/672643/find-the-flaw-breaking-trust-in-your-ai-feature?ajs_uid=207640

What AI feature do you use but not fully trust? Comment below — we might audit it live.

#AITrust #MavenLive #AIAudit #TrustworthyAI #Masto #CranfordTeague

🧵 2/2

NEW COHORT OPENS AUGUST 24th
Designing Trustworthy AI Experiences

If you've been building products for a while, you already have a process that works. AI changes the material that process was built for. Your instincts still hold. Your tools need to catch up.

This course is where you build them.

Six weeks. One real project. A trust lens on every phase.

EXPLORE → https://maven.com/cranford-teague/trustworthy-ai-xd

#TrustworthyAI #AIUX #HumanCenteredAI #AIcourse #Masto #CranfordTeague

NEW COHORT OPENS AUGUST 24th
Designing Trustworthy AI Experiences

If you've been building products for a while, you already have a process that works. AI changes the material that process was built for. Your instincts still hold. Your tools need to catch up.

This course is where you build them.

Six weeks. One real project. A trust lens on every phase.

EXPLORE → https://maven.com/cranford-teague/trustworthy-ai-xd

#TrustworthyAI #AIUX #HumanCenteredAI #AIcourse #Masto #CranfordTeague

NEW COHORT OPENS AUGUST 24th
Designing Trustworthy AI Experiences

If you've been building products for a while, you already have a process that works. AI changes the material that process was built for. Your instincts still hold. Your tools need to catch up.

This course is where you build them.

Six weeks. One real project. A trust lens on every phase.

EXPLORE → https://maven.com/cranford-teague/trustworthy-ai-xd

#TrustworthyAI #AIUX #HumanCenteredAI #AIcourse #Masto #CranfordTeague

EC begins enforcement of AI Act transparency rules

The European Commission’s (EC) AI Office, in collaboration with national authorities, has begun enforcement of the AI Act.…
#Europe #EU #EuropeanCommission #AIOffice #AIsystems #Enforcement #nationalauthorities #TheEuropeanCommission #Transparency #trustworthyAI
https://www.europesays.com/europe/108509/

👋 RC Trust welcomes Marjolein Fokkema.
As Professor of Computational Statistics with Applications in Psychology at TU Dortmund University, she develops interpretable machine learning methods and works to make uncertainty more transparent in behavioral research.
🔗 https://rc-trust.ai/news/news-detail/marjolein-fokkema-joins-rc-trust
#TrustworthyAI #MachineLearning #Statistics #Psychology
Photo: Julian Welz

Designing Trustworthy AI Experiences

Lead AI-augmented projects end to end—from Discover to Deploy—with a repeatable, explainable method.

https://maven.com/cranford-teague/trustworthy-ai-xd

#DigitalTransformation #ExperienceDesign #TrustworthyAI #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

Designing Trustworthy AI Experiences

Lead AI-augmented projects end to end—from Discover to Deploy—with a repeatable, explainable method.

https://maven.com/cranford-teague/trustworthy-ai-xd

#DigitalTransformation #ExperienceDesign #TrustworthyAI #Masto #CranfordTeague

Designing Trustworthy AI Experiences

Setting purpose, boundaries, and guardrails—so AI stays a co-pilot, not a pilot, no matter which technology is under the hood"

Discover More → https://maven.com/cranford-teague/trustworthy-ai-xd

#TrustworthyAI #AIDesign #AIUX #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

Designing Trustworthy AI Experiences

Lead AI-augmented projects end to end—from Discover to Deploy—with a repeatable, explainable method.

https://maven.com/cranford-teague/trustworthy-ai-xd

#DigitalTransformation #ExperienceDesign #TrustworthyAI #Masto #CranfordTeague

Designing Trustworthy AI Experiences

Setting purpose, boundaries, and guardrails—so AI stays a co-pilot, not a pilot, no matter which technology is under the hood"

Discover More → https://maven.com/cranford-teague/trustworthy-ai-xd

#TrustworthyAI #AIDesign #AIUX #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

Communication is compression;
the listener’s assumptions are the decompression algorithm.

Read more → https://createmindfully.substack.com/p/a-sophisticated-assumption-making

#Communication #ArtificialIntelligence #TrustworthyAI #Masto #CranfordTeague

New #publication on Human–#AI Governance and digital sovereignty in the Basque healthcare system. #GenAI should augment—not replace—human expertise, supported by #LivingLabs and democratic data governance.

#OpenAccess https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5931987

#TrustworthyAI #DigitalHealth

A Very Sophisticated Assumption-Making Machine

"You'll hear LLMs (the true name of what we think of as artificial intelligence) called answer machines, reasoning machines, creativity machines. They're none of those things, and the names matter, because your mental model of a tool determines how you use it."

READ → https://www.linkedin.com/pulse/sophisticated-assumption-making-machine-jason-cranford-teague-iysfe/

#TrustworthyAI #DesigningAI #ArtificialIntelligence #Masto #CranfordTeague

A Sophisticated Assumption-Making Machine

"Every compiler, every stylesheet, every command line gave you exactly what you specified and not one pixel more. We built entire disciplines around that fact — programming languages are, at bottom, grammars for leaving nothing to assumptions, because there was nothing on the other end that could make assumptions. The craft was explicitness."

READ → https://createmindfully.substack.com/cp/207956157

#AI #TrustworthyAI #TechEthics #Masto #CranfordTeague

👋 Welcome to RC Trust, Dr. Chris Köcher!
Chris joins Daniel Neider’s group as a Postdoc. His research focuses on automata, formal language theory & verification – and he will continue work on large language models, expressivity and verification. 🤖🔍
#TrustworthyAI #FormalMethods #Verification

👋 Welcome to RC Trust, Paul Fink!
Paul joins Daniel Neider’s group as a PhD student. His interests include formal methods, multi-agent systems, Algorithmic Game Theory, Epistemic Logic and strategic temporal logics. 🔍📐
#FormalMethods #TrustworthyAI #Verification

LG and UNESCO Globally Launch ‘AI Ethics MOOC Project’

SEOUL, South Korea, July 20, 2026 /PRNewswire/ — LG AI Research and UNESCO officially launched the “AI Ethics…
#EuropeSays #Korea #KR #LG #AIdevelopment #AIresearch #AISeoulSummit #ethicalAI #ethicalprinciples #LGGroup #SouthKorea #trustworthyAI #UNESCO
https://www.europesays.com/korea/92482/

CLASS FOR CREATIVE TECHNOLOGISTS:
Designing Trustworthy AI Experiences

AI integrated into YOUR process:
lead AI-augmented projects end to end—from Discover to Deploy—with a repeatable, explainable method.

Sign up and get 50% off → https://maven.com/cranford-teague/trustworthy-ai-xd?promoCode=trust-50

#TrustworthyAI #AIDesign #AIethics #ExplainableAI #Masto #CranfordTeague

Causal discovery under skewed noise. 📊🔎
Daniel Klippert from Alexander Marx’s Causality group at RC Trust presents a poster at ICML 2026 in Seoul. 🎓
His work with Alexander Marx introduces SkewD, a method for causal discovery when real-world data does not follow neat, symmetric patterns. ⚙️🧠
https://rc-trust.ai/news/news-detail/causal-discovery-under-skewed-noise
#ICML2026 #Causality #MachineLearning #Statistics #TrustworthyAI

🤖🔍 Ruhr Fellow David Broczkowski has joined Daniel Neider’s group for a research internship on transformer verification.
He is exploring how AI models can be checked for reliable behaviour when their inputs change – while gaining experience in academic research and practising his German. 🇺🇸🇩🇪
https://rc-trust.ai/news/news-detail/testing-the-limits-of-todays-ai-models
#TrustworthyAI #FormalVerification #MachineLearning

🧪🤖 Can AI make chemical processes safer – and who checks the AI?
Daniel Neider contributes to the 2nd DFG funding phase of FOR 5359 on deep learning for sparse chemical process data.
His subproject focuses on formal verification of neural networks used for anomaly detection in safety-critical chemical processes. 🔍🛡️
https://rc-trust.ai/news/news-detail/checking-the-ai-that-checks-chemical-processes
#TrustworthyAI #MachineLearning #FormalVerification #ChemicalEngineering
Photo: © Schloss Dagstuhl – LZI GmbH, licensed under CC BY-NC-ND

Learning from cause and consequence 🤖🔎
Gabin Agbalé, PhD student in Alexander Marx ’s Causality group at RC Trust, attended the Reinforcement Learning Summer School 2026 in Milan. 🇮🇹🎓
The program connected directly to his work on causal representation learning and offered space for feedback, exchange, and future collaborations. 🌍🤝
https://rc-trust.ai/news/news-detail/learning-from-cause-and-consequence
#ReinforcementLearning #Causality #ArtificialIntelligence #PhDLife #TrustworthyAI
Photo: RLSS 2026/ELLIS Unit Milan

📊 Uncertainty is not a flaw in data science. It is part of making data science trustworthy.
At RC Trust, Dr. Julian Rodemann gave a guest talk on “Uncertainty Quantification and Statistical Inference Under Feedback Loops”.
The talk explored what happens when machine learning models do not only analyse the world, but also influence the data, behaviours, and systems they are meant to understand.
Read more: https://rc-trust.ai/news/news-detail/uncertainty-is-not-a-flaw-but-part-of-trustworthiness
#TrustworthyAI

Research Explorer Ruhr 2026 brought three international early career researchers to RC Trust and into close exchange with their hosts at TU Dortmund University:
Minh Nguyen – Prof. Sven Mayer, Human-AI Interaction
Mohammad Nur Al Homsi – Prof. Daniel Neider, Verification and Formal Guarantees of Machine Learning
Raju Dey – Prof. Markus Pauly, Mathematical statistics and applications in industry
https://rc-trust.ai/news/news-detail/three-visits-three-routes-into-trustworthy-research
#ResearchExplorerRuhr #TUDortmund #TrustworthyAI
Photo: Marquard/RUB

Language models do not only generate text. They increasingly shape how information is found, classified, summarized, and protected.
At #ACL2026 in San Diego, researchers from #RCTrust contributed three papers and one workshop on privacy in NLP, generative web search, and multimodal industry classification.
The contributions involve Ivan Habernal, Bilal Zafar, and Elisabeth Kirsten.
https://rc-trust.ai/news/news-detail/where-language-models-meet-their-blind-spots
#TrustworthyAI #Privacy #GenerativeAI #WebSearch #DataScience
Photo: Elisabeth Kirsten

AI can help researchers write, code, search and check results. But it can also hallucinate sources, obscure processes and challenge scientific trust. 🔍
At the RC Trust Graduate School, @JohannesBreuer discussed what this means for research in the social and behavioral sciences.
https://rc-trust.ai/news/news-detail/when-the-production-of-science-speeds-up-means-for-ensuring-trust-must-catch-up
#TrustworthyAI #OpenScience #ResearchIntegrity #GraduateSchool #RCTrust

WHERE OTHERS LOSE CONTROL – WE STAY IN THE LOOP.

https://blog.infinito.nexus/blog/2026/06/16/where-others-lose-control-we-stay-in-the-loop/

ℹ️ More info: t1p.de/bv5sj See you in Berlin this September! #ResponsibleAI #AutonomousSystems #SummerSchool #BIFOLD #TUBerlin #TrustworthyAI

Summer School

🤖 Hidden instructions shape how AI chatbots behave.
Research from RC Trust by Anna Neumann, Yulu Pi & Jat Singh is featured in the Washington Post 📰
The article explores so-called system prompts and questions of transparency, bias & user control in generative AI.
https://rc-trust.ai/news/news-detail/understanding-the-secret-commands-behind-ai
#AI #ResponsibleAI #TrustworthyAI

Last week at the University of Applied Sciences St. Pölten, the CERTAIN Project hosted a technical workshop and presented at the SAINT Conference 2026.

Our focus: building trust directly into the ML pipeline.

As AI evolves, aligning with the EU AI Act and General Data Protection Regulation is a major challenge. CERTAIN tackles this by turning legal requirements into machine-readable processes.

🔗 https://certain-project.eu/certain-project-progresses-with-strategic-workshop-and-high-impact-presentation-at-saint-conference/

#AI #MLOps #TrustworthyAI #EUAIAct #CERTAINProject

Feel free to reach out if you’d like to connect and chat about these papers or our other work. See you in Barcelona!

#CHI2026 #HCI #AI #TrustworthyAI #SocialMedia #Journalism

(7/7)

Authors: Federico Marcuzzi (INSAIT - Institute for Computer Science, Artificial Intelligence and Technology), Xuefei Ning (Tsinghua University), Roy Schwartz (The Hebrew University of Jerusalem), and Iryna Gurevych (UKP Lab, Technische Universität Darmstadt and ATHENE Center).

See you at #EACL2026 in Rabat 🕌!

#UKPLab #NLProc #ResponsibleAI #Quantization #MLSafety #Fairness #TrustworthyAI #ModelCompression #LLMSafety #EthicalAI #NLP #AIResearch

Work by Nils Dycke & Iryna Gurevych (Ubiquitous Knowledge Processing (UKP) Lab, Technische Universität Darmstadt and National Research Center for Applied Cybersecurity ATHENE)

See you at #EACL2026 in Rabat 🕌!

#UKPLab #LLMs #PeerReview #AIforScience #TrustworthyAI #NLP #Evaluation

🤖📊 How do biases in AI systems change over time?

In a new study in Applied Stochastic Models in Business and Industry, Meltem Aksoy (TU Dortmund University) analyzes thousands of responses generated by different versions of ChatGPT.

The results show: model updates can shift political tendencies – but recognizable patterns remain.

Together with Markus Pauly and the RC Trust, the research highlights why AI systems need continuous statistical monitoring.

#TrustworthyAI #Statistics #DataScience

#NLP #LLMs #MentalHealth #ClinicalNLP #DigitalHealth #ResponsibleAI #NLProc #AIevaluation #ModelEvaluation #TrustworthyAI #Safety #Equity #HumanCenteredAI

We are conducting research to better understand how people from diverse backgrounds perceive and experience Artificial Intelligence (AI) in everyday life, work, and society.

👉 https://docs.google.com/forms/d/e/1FAIpQLSfV3T12Q0XGhJva8KaDae9eIt2PJ0X0rISwa4fH03JQzv-VKQ/viewform

All responses are anonymous and confidential.

#ArtificialIntelligence #AI #Research #TrustworthyAI #DigitalTransformation

📄 New paper published/indexed in @scopus @ieee_bot #AgenticAI

“Anticipatory #AI Governance in Practice: #DataSovereignty, #UrbanAI & #TrustworthyAI #GenAI in the #BasqueCountry”

https://ieeexplore.ieee.org/document/11352770

#OpenAccess @ssrn : https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5660050

#Science4Policy

🔍 We’re hiring 2 Postdocs (m/f/d) in Verification & Formal Methods at TU Dortmund University (Reliable AI Group).

While focused on formal guarantees, we also welcome applications in safe/secure ML and AI quality assurance. 🤖✨

📅 Deadline: March 10, 2026
📩 [email protected]

Join an international, collaborative research environment with access to strong infrastructure & GPU resources.

#Postdoc #FormalMethods #Verification #TrustworthyAI #AIResearch #MachineLearning

Künstliche Intelligenz ist mächtig – aber ist sie auch überprüfbar?
Das neue Forschungsprogramm FIS der Cyberagentur legt den Grundstein für gerichtsfeste, nachvollziehbare KI-Systeme. Drei starke Bietergemeinschaften entwickeln Methoden für die digitale Forensik der Zukunft.
Mehr zur Informationen: https://t1p.de/r2ruf
#Forensik #KI #Cybersicherheit #TrustworthyAI #FIS #DigitaleForensik #XAI