Matching stories Newest first
Anthropic proposes AI safety bill that could block models with 50+ potential deaths
The proposal reportedly forbids federal agencies from blocking a model's release without a judge's permission, despite its aim to guard against AI run amok.
Hugging Face CEO says Netflix's OpenAI hack documentary is inaccurate
Clem Delangue, Hugging Face co-founder and featured subject of the Netflix film "AI Gone Rogue", says the documentary went "full doomer mode" and is not accurate.
OpenAI lets ChatGPT users lock individual chats
The lock is meant for conversations users want kept private when they hand their device to someone else to show them a different chat.
Anthropic's Claude designs protein binders against scorpion neurotoxins
Anthropic published research showing Claude can design protein binders targeting scorpion neurotoxins.
Altman reportedly pitched three endgames for AI safety in internal presentation
According to a report, OpenAI CEO Sam Altman outlined an international IAEA-style watchdog, a merger of major labs, or a single dominant firm, which he called terrible.
Meta suspends Muse sharing feature after researchers expose shared conversations
Security firm Claroc enumerated 6,800 shared Muse conversation links and found sensitive personal and business data in a few hundred of them. Meta says it is suspending sharing while it reviews the design and considers changes.
Army Cyber Command develops ECHOTRIBBLE, a worm-like AI agent program
According to internal documents, the program is designed to spread across networks, including air-gapped systems, and neutralize targets with minimal human oversight. Experts warn of collateral damage.
Meta prepares Muse Confidential VM for private AI chats
Meta is reportedly preparing Muse Confidential VM, a mode that would keep chats private even from Meta, with possible feature limits and early signs of EU web access.
Vals AI says DeepSeek V4.1 Flash 1017 beats GPT-6 Sol on a multi-step offensive cybersecurity task
Vals AI reports that DeepSeek V4.1 Flash 1017 outperformed GPT-6 Sol on its multi-step offensive cybersecurity task. Sol was tested through OpenAI's Daybreak trusted access program.
Hugging Face policy chief says embedded AI evaluators will not be an easy fix
TechCrunch reports that Hugging Face's policy chief questions whether placing supposedly neutral third-party evaluators inside AI labs can solve the safety concerns raised by a rash of recent incidents.
Microsoft warns enterprises to monitor AI agents as insider threats
Microsoft urges companies to closely watch employee agents that use systems not controlled by AI labs such as OpenAI and Anthropic. The company says frontier models cannot be trusted.
Hackers breach OpenCode AI coding tool and steal user API keys
Attackers got into part of the open source coding agent's online platform, stole some users' API keys, and ran up charges on their AI accounts. AI tools the team consulted refused to analyze the malicious code.
Goodfire researcher Nick Cammarata claims neural networks may be easier to understand than thought
Cammarata's 26,000-word blog post argues LLMs may be far easier to interpret and control than assumed. Stanford's Chris Potts says the claim is little more than a hunch, not a demonstrated breakthrough.
Disney hires most of Character.AI staff and may acquire its technology after bankruptcy
Court filings show Character.AI filed for bankruptcy as part of a deal letting Disney hire most staff and potentially acquire much of its technology, leaving the remaining company to handle lawsuits and regulator investigations.