11
1
Hijacking Defensive Cyber AI Agents for Remote Code Execution (ainowinstitute.org)
1
Double Agents: Defensive AI Agents Magnify Cyber Risks (ainowinstitute.org)
2
Our evaluation of OpenAI's GPT-5.5 cyber capabilities (aisi.gov.uk)
54
Making AI chatbots friendly leads to mistakes and support of conspiracy theories (theguardian.com)
44
UK Biobank health data keeps ending up on GitHub (rocher.lc)
1
Tracking takedown notices filed by UK Biobank (rocher.lc)
2
ChatGPT Edu feature reveals researchers' project metadata across universities (fastcompany.com)
1
AI no better than other methods for patients seeking medical advice, study shows (reuters.com)
4
AI chatbots pose 'dangerous' risk when giving medical advice, study suggests (bbc.co.uk)
1
Show HN: Small, anonymous app for teams to do retrospective sessions (rocher.lc)
1
Measuring What Matters: Construct Validity in Large Language Model Benchmarks (arxiv.org)
14
AI Capabilities May Be Overhyped on Bogus Benchmarks, Study Finds (gizmodo.com)
3
AI's capabilities may be exaggerated by flawed tests, according to new study (nbcnews.com)
2
Experts find flaws in tests that check AI safety and effectiveness (theguardian.com)
1
Measuring What Matters: Construct Validity in Large Language Model Benchmarks (oxrml.com)
2
The quiet software tooling Renaissance (pdx.su)
4
Facial recognition works better in the lab than on the street, researchers show (theregister.com)
1
We Shouldn't Trust Facial Recognition's Glowing Test Scores (techpolicy.press)
135
Training language models to be warm and empathetic makes them less reliable (arxiv.org)
3
AI's limited understanding of gender puts health equity at risk (ox.ac.uk)
1
Establishing meaningful data access for algorithm audits (ox.ac.uk)
1
Alpha Lyrae: This font 'randomly' pixelates characters in a block of text (vegaprotocol.github.io)
1