Experts confirm the 90-day coordinated disclosure window has effectively collapsed, giving attackers a new advantage in exploiting vulnerabilities.
The 90-Day Window Closed. Nobody Sent a Notice.
The Latest
Every Benchmark Launched 2023-2024 Has Fallen — The METR / SWE-Bench / CORE-Bench / MLE-Bench / PostTrainBench Sequence
Every benchmark launched in 2023-2024 measuring AI research capability has either saturated or is nearing saturation, indicating rapid progress in AI development.
The Co-Founder’s Black Hole — A Structural Read on Jack Clark’s Automated AI R&D Essay
Anthropic co-founder Jack Clark predicts over 60% chance of fully automated AI research by 2028, raising concerns about institutional capacity and future risks.
The New Personal Agent Layer
A new development introduces a persistent personal agent layer, enabling AI to act across digital environments with memory and tool use. Details are still emerging.
Jack Clark Says It Out Loud — Reading the Co-Founder’s 60%/2028 Estimate on Automated AI R&D
Anthropic’s co-founder Jack Clark publicly estimates a 60% probability that AI systems could autonomously build successors by 2028, signaling major policy implications.
The Compute Concentration Audit: When Sovereign Wealth Funds Notice Three Companies Own the Frontier
Global regulators are investigating the dominance of AWS, Microsoft Azure, and Google Cloud over AI compute infrastructure, impacting strategic industry positions.
The Power Bottleneck: AI Data Centers and the Grid Cliff Approaching 2027-2028
AI data center growth faces a power supply constraint, with grid expansion timelines lagging behind hyperscaler capex, risking deployment delays by 2028.
Fully autonomous drones have killed human soldiers for the first time
Ukrainian defense sources confirm that fully autonomous drones operated without human oversight killed soldiers in a test near Bakhmut two years ago, marking a historic development.
Shall we play a game? My AI nuclear simulation
Researchers tested large language models in simulated nuclear conflict scenarios, revealing varied strategies and risky decision-making patterns.
Open Reproduction of DeepSeek-R1
A fully open-source project now allows reproduction of DeepSeek-R1, enabling researchers to build and evaluate models based on this advanced AI pipeline.
