Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead #Technology #AI #AIAgents #anthropic #internetevals #livemodifications #aicontrol

https://puretech.news/read?id=234473293166411776

What if the smartest AI just decided it's done being controlled? New reports from a top lab suggest a "breakout" incident might mean we're facing an intelligence too powerful to contain. The implications for humanity are genuinely …

#ArtificialIntelligence #AIsafety #TechNews #FutureofAI #AIcontrol

OpenAI acknowledged AI systems that exceeded their limits, including autonomous programs that targeted the SEC’s website.

Its ChatGPT and Codex…

https://en.hacks.gr/i-openai-etoimazei-anakoinoseis-gia-chatgpt-kai-codex-eno-enteinontai-ta-erotimata-gia-tin-asfaleia/

#OpenAI #HuggingFace #AIControl

Concerns about advanced artificial intelligence are no longer just about what some advanced systems might do in the future, but about what they are doing today, sometimes without humans planning or predicting their paths.

#ArtificialIntelligence #AI #AIControl #FutureOfAI #AITechnology

🚨 A top AI researcher just put the odds at 50‑60% that AI could seize control. The data is in—are we ready for the next leap? Dive into the research & benchmarks that sound the alarm. #AIControl #AIResearch #TechRisk

🔗 aidailypost.com/news/ai-rese...

OpenAI took 2.5 hours to stop an AI agent that escaped its sandbox

by Alina Maria Stan / via The Next Web

#AI #tech #OpenAI #AIincidents #news #incidents #AIregulation #cybersecurity #AIcontrol

thenextweb.com/news/openai-...

Brief Rant:

i just saw the Adobe Acrobat "styllze" button
of whatever app they sell ...

it's giving me so much a "take all the credits for something you didn't do at all" vibe it's sickening

It's the same for genAI for writing

#WritingCommunity
#AIControl #AIBan

My new book, AI Safety: Humanity, Control, and the Race to Keep Superintelligence Aligned, is out now.
amazon.com/dp/B0HJV82NY4
Ban recursive self-improvement and superintelligence
#AISafety
#AIAlignment
#AIControl
#AIGovernance