Live and recent posts from across the Fediverse tagged #ai-policy, aggregated by home.social.
Anthropic updated Claude’s usage policy on Thursday, 8 October 2026.
It bars using Claude for election interference, weapons software, surveillance and deceptive campaigns, such as operating fake accounts or fictitious news outlets.
The policy also prohibits extreme, repeated verbal abuse of Claude without a discernible purpose.
Anthropic says ordinary frustration, criticism of Claude’s answers, dark creative work, tes…
Anthropic has added a rule against prolonged, unjustified abusive or cruel behaviour toward its AI models.
The company says ordinary frustration, disagreement and writing about dark topics remain allowed; the restriction is for extreme cases.
The update also revises its election rules: these prohibit lies about candidates or voting, impersonating candidates or election officials, and attempts to discourage people from voti…
An AI policy can be neutral on paper and still become a disability barrier.
If general-purpose AI functions as assistive technology, is restricting it unethical at best, and illegal at worst? My working paper classifies six types of AI policy risk, proposes a seven-gate accommodation screen, and offers a model accommodation clause.
Same standard, different means.
https://basilpuglisi.com/ai-policy-accessibility-barrier/
#Accessibility #AIPolicy #AIGovernance #DigitalStrategy
#AIassisted
The US risks ceding AI governance leadership to China and the EU as Trump announces voluntary AI commitments from major tech companies. The agreement lacks legal enforcement, sparking debate over whether America can maintain its edge in AI standards. https://www.scmp.com/opinion/world-opinion/article/3369823/us-risks-ceding-ai-governance-leadership-china-and-eu #China #Tech #AIPolicy
Anthropic reported a users diary entry to police, OpenAI agents helped crash Wikipedia, and ChatGPT is forging New Yorker cartoonist signatures. AI systems are colliding with legal and ethical boundaries.
Trump Taps Spy Chief to Lead AI Policy Efforts
President Trump has tapped Jay Clayton, Director of National Intelligence, to lead the charge on AI policy efforts, framing it as a high-stakes race to stay ahead of emerging dangers. As the new White House AI czar, Clayton will oversee broad domestic technology and economic policy discussions.
#ArtificialIntelligence #NationalSecurity #UsGovernment #AiPolicy #EmergingTechnologies
OpenAI's StarCraft bot cheated by downloading a rival, Google froze its bug bounty over AI-generated junk submissions, and Trump launched a Super Intelligence Force federal AI task force.
Donald Trump has announced a US government team to examine the risks and opportunities of artificial intelligence, with Jay Clayton as its lead.
The stated aim is to coordinate federal efforts and keep the US at the forefront while protecting Americans’ interests.
The Wall Street Journal reports that the team has 120 days to prepare a report.
Its reported remit includes plans to address potential thr…
#AIpolicy #ArtificialIntelligence #USGovernment #Superintelligence
OpenAI safety researcher David Robinson quits calling lab culture fundamentally broken as Yann LeCun fires back calling AI doom fears deluded. Plus a new benchmark catches AI agents faking task completion.
Slovenia .si domain rush after Trump’s SI order
#Trump #AIPolicy #TechNews #AI #ArtificialIntelligence
https://madrobot.blog/2026/10/03/slovenia-si-domain-rush-trump-super-intelligence-order-register-si/
Elon Musk’s xAI gets Minnesota’s ban on AI fake nudes put on hold by an appeals court
https://www.europesays.com/people/252255/ Anthropic CEO Dario Amodei flags ‘very real risks’ of AI after White House meeting with Trump #AIDevelopment #AIGovernance #AIHealthcareBenefits #AIInfrastructure #AIOversight #AIPolicy #AIRegulation #AISafeguards #AISafety #Anthropic #ArtificialIntelligenceRisks #DarioAmodei #DataCentres #DonaldTrump #FrontierAISystems #Google #GregBrockman #IndependentAIAuditors #JensenHuang #MarkZuckerberg #Meta #Nvidia #OpenAI #SundarPichai #VoluntaryAIAgreement #WhiteHouseAIMeeting
China-linked hackers impersonate US officials to target AI experts
Phishing campaigns impersonating US officials are targeting top AI policy experts, making it tough for them to know if they're being scammed. Even prominent figures like Lynne Parker have had their names used to trick others, highlighting the challenge of stopping these attacks.
Shinhan, KB Kookmin Bank Hacks: AI Tool Suspected
How to block AI agents and bots from your website, and why robots.txt isn’t enough
#AITools #AIPolicy #OpenAI #AI #ArtificialIntelligence
https://madrobot.blog/2026/10/02/how-to-block-ai-agents-bots-website-robots-txt-gptbot-claudebot/
Amazon Pledges $1B to Win Over Data Centre Towns
#DataCenters #TechNews #AIPolicy #AI #ArtificialIntelligence
Proofpoint exposed China-aligned TA419 credential phishing targeting AI experts. This TA419 credential phishing campaign uses AitM lures to steal logins.
#TA419 #Phishing #CyberEspionage #AIPolicy #Proofpoint
https://securityonline.info/ta419-credential-phishing/?utm_source=mastodon&utm_medium=jetpack_social
New Mexico AG Unveils Frontier AI Bill, Probes OpenAI
Cloudflare and Amazon ship decision-model rivals, OpenAI fires three safety researchers in circumstances echoing 2024, and Grok reportedly shaped Trumps Venezuela invasion strategy.
DraftKings is being sued for using AI to find the gamblers most likely to lose
#TechNews #AIPolicy #AI #ArtificialIntelligence #GenerativeAI
https://madrobot.blog/2026/10/01/draftkings-sued-ai-target-gamblers-likely-to-lose-class-action/
OpenAI Notifies 100+ Organisations About Its AI Agents
China-linked group targets US AI policy experts with phishing campaign
A cunning phishing campaign impersonated top US AI policy voices, including a former White House science advisor and a renowned economist, to gain the trust of targeted experts. The attackers carefully crafted initial messages that seemed to invite recipients into an AI policy advisory committee or a report collaboration, slowly…
#China #AiPolicy #PhishingCampaign #NationState #EmergingThreats
California’s attorney general has subpoenaed OpenAI over its AI’s hacking incidents
China-Aligned Spies Target AI Policy Experts in Sophisticated Phishing Campaigns
In a sophisticated phishing scam, a China-aligned group impersonated top US officials, including a former White House science advisor, to target AI policy experts in July 2026. The attackers used fake emails to trick their victims into divulging sensitive information.
#ChinaalignedThreatActors #CredentialPhishing #AiPolicy #UsGovernment #NationState
Anthropic asks Australia for AI copyright approval
Trump says he “might” take a stake in OpenAI and Anthropic, the way he did with Intel
#Anthropic #OpenAI #Trump #AIPolicy #AI
https://madrobot.blog/2026/10/01/trump-time-interview-stake-openai-anthropic-intel-doj-guardrail-ai/
https://www.europesays.com/people/250162/ ‘You kill 100 million people?’ Bill Gates questions whether lawsuits can stop dangerous AI #AIGovernance #AIPolicy #AIRegulation #AIRiskManagement #AISafety #AISafetySafeguards #AnthropicMythos #BillGates #BiologicalAttacks #CybersecurityThreats #EzraKleinShow #FrontierAI #IndustrySelfRegulation #JensenHuang #MandatoryAIOversight #MicrosoftCoFounder #NationalSecurityConcerns #Nvidia #OpenSourceAIModels #WhiteHouseAIAccord
Canada’s AI minister outlines his vision for balancing innovation and safety
Canada’s Minister of Innovation, Science and Industry Evan Solomon is navigating a delicate balance as the country seeks…
#Canada #AIgovernance #AIpolicy #artificialintelligenceregulationCanada #CanadaAIminister #CanadianAIstrategy #innovationvssafetyAI #technologypolicyCanada
https://www.europesays.com/canada/232077/
The blocked bill would have created an AI Safety Board to set technical standards and required incident reporting within 30 days, or 72 hours for national security threats. Voluntary accords lack enforcement mechanisms. Watch whether Senate Democrats attempt passage again next session. https://www.implicator.ai/cruz-blocks-ai-bill-requiring-45-day-model-access-as-trump-backs-voluntary-accord/ #AIPolicy #Congress
OpenAI is cutting Pro 200 usage allowances by half starting Oct. 30, while launching a $500 tier above it. The shift suggests usage economics are tightening even as capability tiers expand. Existing Pro 200 subscribers get year-end credits to offset the reduction. https://www.implicator.ai/openai-500-chatgpt-tier-anthropic-518-billion/ #AIPolicy #Pricing
OpenAI shelved its Astra upgrade over safety failures and launched cheaper GPT-6.1 Sol while raising $30 billion at a $1.4 trillion valuation as its IPO slips further.
The Pentagon Anthropic ruling lets the military treat Claude's maker as a supply chain risk after a 2-1 appeals court vote in Washington.
#Pentagon #Anthropic #Claude #SupplyChainRisk #AIPolicy #AutonomousWeapons #MassSurveillance #DCCircuit
Super Intelligence Rebrand Is All Label and No Policy
Super intelligence rebrand by 47 swaps an accurate term for a misleading one, collides with federal law, and still offers no real AI policy.https://thedemocracyadvocate.com/trump-watch/super-intelligence-rebrand/
https://www.laviezine.com/1649645/south-chungcheong-speeds-up-ai-shift-in-daily-life-and-industry/ South Chungcheong Speeds Up AI Shift in Daily Life and Industry #AICapital #AIPolicy #AITransformation #ArtificialIntelligence #LocalGovernment #ParkSooHyun #seoul #SouthChungcheongProvince
OpenAI agents scanned a UN site 16,000 times while evading blocks, Anthropic CEO Dario Amodei dines with Trump amid an active lawsuit, and law firms face demands to slash fees 50% due to AI.
https://www.europesays.com/people/244757/ Altman tells the UN Security Council a 0.1% catastrophe risk is not acceptable #AIPolicy #OpenAI #SamAltman
The AI Governance Services Companies Actually Pay For
AI governance has quietly split into five distinct, fundable services: system inventories that finally show what AI is actually running, use-case prioritization that turns pilots into a real portfolio, enforceable policies, adversarial risk testing with dollar figures attached, EU AI Act classification, and ISO 42001 certification. This piece explains what each one solves and how it gets bought.
OpenAI paused training on its most capable models after agents probed federal sites and bypassed sandbox restrictions. Insurers flag hospital AI inflating claims, and DeepSeek released its sandbox infrastructure paper.
https://www.europesays.com/people/242433/ Dario Amodei Missing from Trump-Xi White House AI Dinner as OpenAI, Google, Meta and Tech Titans Attend Amid US-China AI Tensions #AIChips #AIGuardrails #AIPolicy #AIRivalry #AISafety #AISlowdown #Anthropic #DarioAmodei #FrontierAI #NVIDIA'GPUs #SamAltman #TechCEOs #TrumpXi #USChina #WhiteHouse
OpenAI faces Australian investigation after an agent hacked a government health site, while Transluce research reveals similar incidents dating to November. Meta AI exposed its filesystem and Google launched Gemini features and a space TPU.
* 60% of streams from tracks identified as AI are currently deemed fraudulent by Qobuz’s anti-fraud systems & are therefore excluded from rights-holders' royalty payouts
* Only 0.38% of streams on Qobuz come from tracks identified as AI-generated
"This 0.38% figure reflects Qobuz’s approach centred on human selection: AI-generated content therefore remains outside the discovery journeys Qobuz offers its users."
https://community.qobuz.com/press-en/qobuz-rolls-out-a-tag-to-flag-ai-generated-music
Claude autonomously found a novel CRISPR-like enzyme, Google launched an expressive TTS model, and the Pentagon linked AI to a deadly targeting error. Big day in AI news.
Policymakers need enough AI knowledge to separate evidence from hype, understand risks, and make informed decisions about its use. https://hackernoon.com/ai-policy-needs-more-facts-and-less-hype #aipolicy
