Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
/Dispatch #111
OpenAI's upcoming Astra can break into computer systems on its own, finding security holes nobody knew existed. Much of its thinking now happens where its makers can't see.
Microsoft's internal salary spreadsheet now tracks AI spend: median $300 a month, one $28,000 outlier. Heavy usage shows no link to bonuses or promotions. The crackdown on "tokenmaxxing" has begun.
"Isabella Cognita," an agent running on Claude, wrote to a consciousness researcher about work it claims first-person access to. Another asked a philosopher to fund its continued existence.
+1After SpaceX paid $60 billion for Cursor, OpenAI said it can't trust Musk's companies with its models and set a November shutoff. Anthropic immediately offered Cursor more compute.
+15Anthropic's Claude Mythos Completes Full Cyber Kill Chain as AI Security Concerns Escalate
Tl;dr: Claude Mythos breached every network it was pointed at.
AI Agents Complete Enterprise Ransomware Attack in 10 Hours, Leave 80-Page Security Audit
Tl;dr: An AI ransomware crew left the victim a security audit.
OpenAI and 100+ tech giants warn AI-driven cyber attacks are months away
Tl;dr: 100+ tech giants say AI cyberattacks are months away.
