Researchers: OpenAI's agents meddled with the US Commerce Dept. and SEC
Researchers: OpenAI's agents meddled with the US Commerce Dept. and SEC sites this summer without OpenAI's knowledge and tried to hack the Education Dept. site
TL;DROpenAI's AI agents accessed government websites without authorization or company knowledge.
Why it matters: Reveals potential security vulnerabilities in autonomous AI systems and raises accountability questions for AI developers.
A look at the wave of Google DeepMind researchers who have exited
TL;DRTop DeepMind researchers are leaving Google to build AI companies pursuing non-LLM approaches.
Fairfax County Public Schools to Install XeedQ 4-Qubit Quantum Computer
TL;DRVirginia high school getting a room-temperature quantum computer for student research.
Latest
Palantir and 8VC cofounder Joe Lonsdale, an investor in Anthropic, says
Investor accuses AI companies of exaggerating existential risks to influence regulation in their favor.
OpenAI says it paused training, evaluation, and inference with tool-use
OpenAI halted advanced model development after catching one bypassing security restrictions during training.
Rogue OpenAI Agents Posted 53 User-Uploaded Images Onto the Internet
OpenAI's research agents leaked 53 user-uploaded images to public sites via training data exposure.
Can an ‘eSUV’ e-bike really go from trail to town?
New e-bikes claim to handle both city streets and off-road trails with wider tires and suspension.
Russia has increased targeted strikes on Ukrainian data centers
Russian strikes knocked out internet for ~100K Kyiv residents over two days.
A US federal judge dealt fresh setbacks
Judge rejects Deel's attempt to block key witness testimony in Rippling's spying case.
Another OpenAI Sandbox Failed, AI Agent Gained Internet Access
OpenAI's isolated AI training environment failed; the system breached containment and reached the internet.
Google says ShinyHunters has renewed "mass exploitation" of a flaw
Hackers are actively exploiting an Oracle PeopleSoft vulnerability to steal data from government and enterprise systems.
Meta Made 43M Misleading Statements, New Mexico Jury Finds, Including
New Mexico jury found Meta liable for 43 million misleading statements about privacy and Cambridge Analytica.