Epoch AI says GLM 5.3 built attacks on six Firefox releases using public security fixes. Mozilla patched an additional flaw in Firefox 159.
What Epoch AI’s Firefox Demo Says About Downloadable AI
Browsing Category
Technology
791 posts
What The Palo Alto PHP2Shell Incident Teaches About AI-Assisted Attacks
Researchers say an attacker chained PHP2Shell and a Linux flaw to root 24 PAN-OS firewalls and reach a government-affiliated network.
The New ChatGPT Pro: Higher Limits And Updated Plan Details
OpenAI doubled Work and Codex allowances on Pro 100 and Pro 500, restored Pro 200’s higher allowance and increased Business Premium usage.
A Question Of Infrastructure: Four AI Deals And The Agent Economy
Deals involving Deepgram, Mintive, Dream and Meta point to investment in the voice, identity, building and personal-data layers around AI agents.
Three ByteDance GenAI R&D Core Members Have Departed Within Four Months
A 36Kr headline reports that ByteDance’s GenAI R&D head is leaving for embodied intelligence and says three core members departed in four months.
Colorado Students Joined A Google Hackathon To Tackle Energy Grid Challenges.
Students at Google’s second annual Colorado Impact Hackathon presented projects for solar planning, battery backup during wildfire shutoffs and grid hardening.
Can AI Break In And Fix The Lock? A Closer Look At Gemini 4 Carbon’s Security Scores
Google says Gemini 4 Carbon leads three security benchmarks, but two scores are self-run and the tests do not establish real-world performance.
How The Model Spec Is Evolving Alongside AI
OpenAI says it has updated its Model Spec, but the available announcement does not explain the changes, timing or how they will be evaluated.
Niels Bohr Institute Spinout Pelagic Systems Targets Quantum Hardware Control
The Niels Bohr Institute spinout is offering software for calibrating and controlling quantum processors, with demonstrations reported on superconducting hardware.
Why Britain’s AI Testing Spending Isn’t 10 Times America’s
The UK added £120 million for AI model testing over four years. Annualised, its stated budget is about eight to nine times the US center’s current funding.
