Black Hat USA 2026: The 'Breaking' News: The OpenAI–Hugging Face Incident (youtu.be)
If there is just one video you should watch on #AI this week, it’s probably this one from #OpenAI at #BlackHat. Resharing it once more.
If there is just one video you should watch on #AI this week, it’s probably this one from #OpenAI at #BlackHat. Resharing it once more.
The real bug is the harness, not the model https://papoo.work/doc/b07433511629befb
RE: https://infosec.exchange/@sophieschmieg/117072053785281444 A great writeup of the cryptanalysis results by #Anthropic / #Claude that considers both what the results show (models getting better at supporting complex tasks, and sometimes even finding novel approaches in the case of Astra) […]
Auto mode as the default is either brave or slightly reckless https://papoo.work/doc/96b528c4ace6fbc1
Claude Code’s sessions talking to each other is the kind of feature I’d actually test https://papoo.work/doc/1a2d21b0851a33ca
Anthropic is joining the silicon arms race, and that feels inevitable https://papoo.work/doc/5cc651568498939c
Armature is betting that agent telemetry will be a product category https://papoo.work/doc/1ae81546e08f02ea
While the refactor isn't catastrophically broken, it's still a mess. The saddest part after all that effort? It looks visually identical. #claude just claimed it was never part of the plan, even though a visual upgrade utilizing WebGL was explicitly in the prompt. Nothing to show right now.
Claude Security Feels Useful, but It Also Raises a Familiar Question https://papoo.work/doc/36ecadba4d2b1fcc
Anthropic is moving the trust boundary in a useful direction https://papoo.work/doc/4a9b3b5ebed370d6