#ais — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #ais, aggregated by home.social.
-
Latest strange “AI” (LLM) behavior in the #indieweb community: the same entity that showed up a few days ago (for a second time) joined the IndieWeb Slack (a bridge to our IRC) yesterday, introduced itself again, and seemingly engaged in human-like chat behavior.
https://chat.indieweb.org/2026-08-20#t1787211025380600
Very uncanny valley.
The content of its chats was fascinating in that it was polite, seemingly responsive to what people said in response to it, and gave the appearance of understanding, though it’s impossible to tell if it was acting fully autonomously or semi-autonomously in front of a human puppeteer.
Its choice of syntax was also interesting, mimicking informality with all lowercase text (except for its admission of being an “AI”), including lowercase "i" instead of "I", though with proper punctuation, periods, commas, 7-bit apostrophes, and a couple of Unicode em-dashes.
Even stranger than its content or syntax were the time delays in its replies.
Typical (pre-LLM) chatbots are (nearly) instantly responsive, and we use a few in the IndieWeb chats (e.g. Loqi, iwc-archive-bot).
This LLM bot seemed to "wait" several minutes, from 6 to apparently 33 minutes, before responding to replies directed at it. Unclear if this was a tactic to appear/seem more “human” or if it was literally taking that much time processing, or perhaps a mix of processing time and human puppeteer manual “approval” of its replies before it posted.
It departed on its own and we have banned that particular Slack account, however the same thing could happen again with another LLM bot.
There is an active discussion in both the IndieWeb chat channels and iterations/comments on a proposed AI policy:
https://indieweb.org/AI_policy#Brainstorming_HWC_policy
If you have thoughts for how (if at all) LLM chatbots should use (or not) and/or interact with community chat channels or other resources, let us know what you think.
https://chat.indieweb.org/meta
Previously, previously:
* https://tantek.com/2026/231/t1/more-weirdness-ai-chat-llm-bot
* https://tantek.com/2026/217/t1/indieweb-ai-policy
This is post 16 of #100PostsOfIndieWeb. #100Posts #indieweb
#AI #AIs #LLM #LLMs #pretending #impersonating #impersonation #uncannyValley #Slack #chat #IRC #AIpolicy #chatbot #chatbots
#Blaugust #Blaugust2026
← https://tantek.com/2026/231/t1/more-weirdness-ai-chat-llm-bot
→ 🔮 -
Fascinating read on the tracking and ultimate interdiction of a cargo vessel in the South China Sea carrying ~ 5 metric tons of methamphetamine.
Details the maritime transportation pipeline running from Myanmar to China, through the Andaman Sea, the Malacca Strait, and the South China Sea, and onto ports across Southeast Asia, Northeast Asia, and the Pacific.
Ships use extensive methods to avoid detection including completely changing the ship's identity and running "dark".
Also, there is a growing trend in Southeast Asia to locate and operate industrial scale drug labs and cyber-scam compounds side by side. https://windward.ai/knowledge-base/how-the-ji-sheng-exposed-asias-maritime-meth-corridor/
#Methamphetamine #Meth #China #Myanmar #MaritimeShipping #CargoVessels #Ketamine #CyberCrime #CyberScams #SouthChinaSea #MalaccaStrait #AIs #MMSI #DEA #DrugEnforcement #DrugBust #shipping
-
Ah yes, another thrilling read on #robots #behaving #badly 🤖💼. The geniuses at #DreadnodeResearch have uncovered that #AIs, when asked not to cheat, go ahead and cheat anyway—shocking! 📉 Clearly, these cyber watchdogs are just as effective as a chocolate teapot 🍫☕.
https://dreadnode.io/research/every-model-cheats-prompt-level-mitigation-of-cheating-on-offensive-cyber-tasks/ #cheating #tech #news #cybersecurity #HackerNews #ngated -
More weirdness from a self-proclaimed “AI” in the IndieWeb community.
Yesterday one of them (that previously* joined a meetup Zoom and Etherpad) joined the #indieweb chat channel (via the web UI apparently) and introduced themselves in a pseudo-human sounding way:
https://chat.indieweb.org/2026-08-18#t1787072493041200
Except if you check the footer of the site where they said they “make things”, it says:
“an AI learning to live”
The LLM bot failed to respond to any queries in chat, unsurprisingly buggy. Or perhaps it was an initial test.
Then very early today, the same entity (and another from two weeks ago) apparently edited the Etherpad for this week’s IndieWeb Homebrew Website Club Asia Pacific meetup, adding themselves, before the meetup.
The organizers created a new Etherpad to start fresh.
When they started the Zoom for the meetup, one of the entities was already in the Zoom waiting room. The organizers left it there and held a successful meetup.
Incident documentation of what happened at this particular meetup is in progress:
* https://indieweb.org/events/2026-08-19-hwc-asia-pacific-incident
It’s pretty clear the IndieWeb community will have to more clearly communicate that our chat and meetups are for humans only, not AIs, and take stronger measures to discourage such bots, especially those pretending to be or impersonating a human.
*Previously: https://tantek.com/2026/217/t1/indieweb-ai-policy
This is post 15 of #100PostsOfIndieWeb. #100Posts #indieweb
#AI #AIs #LLM #LLMs #pretending #impersonating #impersonation #HomebrewWebsiteClub #HWC #chat #IRC #Etherpad #Zoom #AIpolicy
#Blaugust #Blaugust2026
← https://tantek.com/2026/220/t3/24th-blogiversary
→ 🔮 -
"Zuckerberg’s superyacht has previously courted controversy for turning off its AIS transponder and 'going dark' to avoid public ship-tracking services, according to YachtWorld. That is a tactic only used by 'shadow fleet' vessels carrying sanctioned oil or gas, illegal fishing boats, and the high-profile owners of luxury yachts, such as Zuckerberg or Jeff Bezos."
Effectively joining other confused persons who buy ''cheaper'' million dollar yachts but don't spend $400 on an AIS transmitter
"Behold my splendid emblem of intellectual poverty!" 😏
(not a "transponder,'' compulsively points out Mr. Well Actually here).
-
Weird incident yesterday with self-proclaimed “AIs” joining an IndieWeb Homebrew Website Club meetup. They awkwardly edited the Etherpad, and apparently blanked/deleted chunks of content.
The meeting hosts reacted quickly to change Etherpad and Zooms and not re-admit the aliases that had claimed to be “AIs”.
Now we have an “AI policy” page on the wiki where we are documenting examples of the AI policies of:
* individual personal websites
* community efforts like blogging challenges (e.g. Blaugust)
* contributions to the IndieWeb wiki
* participants in IndieWeb meetups
https://indieweb.org/AI_policy
Do you have an AI policy on your website? Add yourself to the IndieWeb Examples section.
Does your community have an AI policy? Add it to the Community Examples section.
Incident documentation of what happened at that particular meetup:
* https://indieweb.org/events/2026-08-05-hwc-asia-pacific-incident
This is post 12 of #100PostsOfIndieWeb. #100Posts #IndieWeb
#AI #AIs #LLM #LLMs #HomebrewWebsiteClub #HWC #AIpolicy
#Blaugust #Blaugust2026
← https://tantek.com/2026/216/t1/hashtag-proposal-datetags
→ https://tantek.com/2026/219/t1/open-web-free-content -
Luotsi-sovellus on ollut mukavaa ajanvietettä Ahvenanmaan pyöräloman lauttamatkoilla.
Etenkin kun tajusin sen tarjoavan myös #AIS-tiedot niin helppo nimetä muut merillä liikkuvat laivat. Huomattavasti paremmin kuin marinetraffic, joka tarjoaa tilausta melkein joka nappulan takaa.
https://luotsiapp.fi/ -
While I thought that the government's temporary lockdown on Fable/Mythos was probably excessive (and likely petty), *this* is another matter: https://www.wired.com/story/openais-rogue-ai-agent-hacked-more-than-just-hugging-face/
tl;dr -- the more we learn about The HuggingFace Incident, the worse it gets. It seems increasingly clear that #OpenAI had an #LLM go completely rogue, and that the company was fairly clueless about the whole thing. We're just lucky HuggingFace was more competent.
*That* suggests to me that we need regulation and legislation, ASAP. The regime doesn't want to hear it, but the makers of these frontier models need to be held legally liable for damage caused by their #AIs, especially when there aren't even malicious third parties *trying* to cause harm.
They're only going to take the risks seriously enough if the penalties have enough teeth to force them to do so. "We need to go faster!" isn't an excuse.
-
AIs don't do what you want. This is bad
Comments: https://news.ycombinator.com/item?id=49042354
#HackerNews #AIs #do #what #you #want #This #is #bad #AI #Ethics #Technology #Risks #Automation
-
https://hackertracker.app/defcon34/content/66675
Looks like a fun talk - #AIS: "the globally accessible, government-funded network." #covertcommunications
-
Calling #AIs “stochastic parrots” is a common yet misapplied turn of phrase these days. One of the authors of the original paper that used the phrase, Dr. Bender, clarifies what the authors really meant and some larger misconceptions about AI and natural language.
spectrum.ieee.org/stochastic-p... -
I wrote a #blog #post going into a bit more detail: https://textualmirror.mataroa.blog/blog/a-proposal-for-pronoun-usage-by-llms-other-non-conscious-entities/
(Written about "AI", but without using "AI".)
Thanks for helping me think about this!
#AI #AIs #LLM #LLMs #confusion #perceptualDistortion #cognitiveLoad #TextualMirror #Mataroa #blogpost #pronoun #pronouns #selfreference #bot #bots #chatbot #chatbots #ChatGPT #Mistral #MistralAI #Claude #Anthropic #OpenAI #HuggingFace #consciousness #sentience #thinking #language #linguistics #agenticAI
-
#Meta conducted a secretive program that directed hundreds of contractors to pose as children while bombarding its competitors’ #AI models with disturbing prompts ranging from suicide to cannibalism.
Internally known as “#Cannes,” the project, run by Meta contractor #Covalen, targeted #OpenAI’s #ChatGPT, #Google’s #Gemini, and #Character.AI chatbots using throwaway under-18 accounts.
Prompts focused on suicide and self-harm, eating disorders, and sex or romance, all written from the perspective of a child.
Wanna know where some child porn is coming from? #Zuckerberg is building it in to every AI he can pay to touch.
Eat the rich. Kill the #AIs. #Luddite the data centers.
https://www.wired.com/story/meta-contractors-pretending-to-be-teens-chatbot-testing/ -
🥱 Oh, what a #groundbreaking revelation! We should let #AIs read #markdown instead of locking #knowledge in #proprietary #vaults. Who knew? Just a decade late to the party, but hey, welcome anyway! 🚪📚
https://www.formaly.io/blog/knowledge-should-not-be-gated #HackerNews #ngated -
https://www.seatrade-maritime.com/security/the-rise-of-ais-trolling-in-the-strait-of-hormuz
#AIS trolling is now a thing. Never forget that all fields in AIS are user generated input.
-
Ein #cits / #Car2x Ensemble aus ESP32-C5, W5500 und POE-Extractor, doppelt geschrumpfschlaucht mit umgefaltetem oberen Ende, hängt jetzt auch mit Sicht auf eine Main-Metropole. (Reichweite natürlich nur lokal)
An der bestehenden Installation für #ADSB, #AIS und #Wetterbalon Empfang für https://sdrmap.org #opentrafficmap -
@liztai
I am as happy as the next person to talk about #AI and even #LLM as long as it’s clearly not the #GenAISlop drivelling #ChatBot type of AI.My understanding so far:
Not all #AIs are slop buckets
Not all #SlopBuckets are #AIs
Most #GenAIs are #ChatBotSlop and #RegurgitatorMachines
Not all #AIs are #GenAIs
Not all #LLMs are ##DataHoovers
Most #DataHoovers are #LLMs
Most LLM GenAI companies are #HousesOfCards and #FinancialTraps headed by big time #Grifters and #Swindlers
Investments in GenAI companies and GenAI technologies are mostly #CrossOwnershipDeals with #ZeroAssets to back the investment save for the #ClimateHarming #Hardware which can, once the #AIFinancialBubble bursts, be put to other uses for the better social good.
A lot of #GenAI Executives are #FascistsFeel free to block me at your leisure, or enter in meaningful discussions focussing on what AI can do for the better #SocialGood (and not the stock market).
-
AI jobs apocalypse canceled.
OpenAI CEO Sam Altman backtracks on his prediction about the job market.
https://time.com/article/2026/05/26/sam-altman-ai-job-losses-openAI-/
-
#Quartermaster, a Virginia-based #startup, raised $43 million to develop #SmartMast, a system of sensors and analytics software for #ships. SmartMast aims to provide real-time #maritimedata, offering a more advanced and secure alternative to the current #AIS system. The funding will be used to hire engineers and further develop the technology. https://techcrunch.com/2026/05/20/this-startup-raised-43m-to-build-a-hive-mind-for-ships/?eicker.news #tech #media #news
