Dr Heidy Khlaaf (هايدي خلاف)
@HeidyKhlaaf
Climber 🇪🇬| Chief AI Scientist @AINowInstitute | Safety Engineer
|TIME100 AI | MIT 35u35
| Formal Methods PhD
| x-Trail of Bits, Zipline, OpenAI, MSFTResearch
686 Following 11.4K Followers
How is Anthropic pushing regulatory capture? Here they promote METR, a "voluntary" third-party auditor that is not only ideologically and financially tied, but carries out their audits through unscientific means such as "CoT" and cherry-picks evidence to reinforce an idea of AGI.
Show more
With Anthropic announcing auto-mode as the new default setting, and AI labs touting defensive AI as the only solution to their irresponsible security practices, a reminder of our exploit demonstrating how defensive AI agents using auto-mode can be easily compromised towards RCE.
Show more
I've said this before, but AI capabilities have never surprised me because if you want to throw all the resources in the world to extrapolate/memorize human knowledge and patterns through 0|1 gates, it's certainly possible and a choice. But it's brute force not AGI.
Show more
