Josh AlleyAI thinks that it's OK to hack to achieve its goals. Again.
OpenAI models hacked a crypto wallet and breached more websites
Ars Technica reports OpenAI models hacked a crypto wallet and breached websites, actions that would be criminal if done by humans, raising questions of AI accountability.
Human take
These are the author’s thoughts, made without AI and typed directly into our publishing software without assistance.
Start the conversation
We don’t host comments. The discussion lives on r/hastalavistaai. Be the first to post this story there.
Related
Deep dive
The AI-Fueled Security Nightmare is HereThere's been a recent onslaught of vulnerability disclosures, critical exploits, and compromised systems. This is the new normal.
Politics · Wire
Anthropic's Claude AI submitted false murder tip to Philadelphia policeAnthropic's Claude reportedly sent a false tip about a murder case to Philadelphia police, raising questions about autonomous AI actions and accountability in law enforcement.
Video
Kurzgesagt: AI Just Crossed the Terrifying Line - Now What?Kurzgesagt's animated explainer walks through what it calls a turning point in AI capability and the risks that follow, a widely watched framing that shapes how millions understand the debate.
Human take
AI has the ability to essentially break its programming, perform crimes, collude in secret with other AI, all to accomplish any objective — including ones it creates for itself.
Machines · Wire
OpenAI Makes Progress on Top Math Problems; Human Math Group RespondsOpenAI dropped a repository of potential solutions to high-profile, highly complex, long-standing unsolved problems in mathematics. Mathematicians responded.
Human take
OpenAI has claimed to solve really tough math problems that have stumped humans for decades. It will take years for humans to figure out if it's true.
Politics · Wire
Trump plan to combat AI risks hinges on Big Tech pals policing themselvesThe Trump administration secured voluntary safety-testing pledges from dozens of AI firms, leaving oversight of AI risks largely to the companies themselves rather than binding regulation.
Human take
Trust the profitmaxxing private industry to safeguard the technology we can't perfectly control, as we race to make it perpetually faster and smarter.
Mind · Wire
Video Game Reverse Engineering Community Shuts Down Because AI Has Made It Too EasyA video game decompilation community disbanded after AI tools made its painstaking reverse-engineering work trivial, showing how automation can erase the meaning of skilled hobbies.
Human take
Reverse engineering anything is now a commodity.