Don't expect quick fixes in 'red-teaming' of AI models. Security was an afterthought

Some 2,200 DefCon competitors tapped on laptops seeking to expose flaws in eight leading large-language models representative of technology’s next big thing. But don’t expect quick results from this first-ever independent “red-teaming” of multiple models.

Related Keywords

Cambridge ,Cambridgeshire ,United Kingdom ,Boston ,Massachusetts ,United States ,Switzerland ,Zurich ,Züsz ,White House ,District Of Columbia ,Swiss ,Carnegie Mellon ,Ram Shankar Siva Kumar ,Andreww Moore ,Google Bard ,Bruce Schneier ,Ross Anderson ,Hyrum Anderson ,Tom Bonner ,Las Vegas ,Michael Sellitto ,Gary Mcgraw ,Samsung ,Cambridge University ,Berryville Institute Of Machine ,Microsoft ,Swiss University ,Google ,Us National Security Commission On Artificial Intelligence ,Verizon ,Silicon Valley ,Berryville Institute ,Security Commission ,Artificial Intelligence ,Florian Tram ,Snot With ,Bug But With ,Ai ,Technology ,Internet ,