What happened
A METR study found AI has dramatically accelerated reported cyber vulnerabilities in 2026 versus 2025, citing projects like cURL, OpenSSL, Firefox, and Microsoft.
Why it matters
Acceleration is uneven—AI shows minor gains in mathematics (e.g., solved Jacobian conjecture) but no measurable acceleration in AI research across key benchmarks. This suggests AI's impact is concentrated in specific areas, not uniform.
What to watch
Whether AI triggers similar 'phase changes' in other science and tech fields, as seen in coding (2025) and cyber (2026).
Summaries like this, in your inbox every morning.
The METR study highlights differential acceleration: AI contributes significantly to cyber, marginally to math, and not measurably to AI research. The author suggests acceleration occurs when models undergo a 'phase change' for a skill, as seen with coding in 2025 and cyber in 2026. This raises a critical question: will similar phase changes occur in other fields? The findings imply AI's impact isn't uniform, so stakeholders should anticipate concentrated gains rather than broad-based progress.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
