Curated AI news from the last 24 hours — models, agents, enterprise, and developer tools.
OpenAI announced that an internal version of its upcoming Astra model solved ten previously open problems in mathematics and theoretical computer science, including the existence of non-sofic groups and new sphere-packing bounds. The results were published as machine-checkable Lean proofs on GitHub for roughly $2,000 in compute, and Fields Medalist Timothy Gowers said he would recommend one of the proofs for a top journal without hesitation.
openai.com ↗AI-content detection startup Pangram raised $9 million as demand grows for tools that flag AI-generated text and images online. The company launched Pangram 4, its next-generation text detection model, alongside a new AI image detection model.
techcrunch.com ↗LinkedIn introduced a feature letting users flag posts that look like low-quality AI-generated content, as the platform contends with a rise in "AI slop." The company is also replacing its own AI writing assistant feature with a proofreading tool.
techcrunch.com ↗GitHub deprecated Gemini 2.5 Pro and Gemini 3 Flash across all GitHub Copilot experiences as of July 31, 2026, suggesting Gemini 3 Flash users migrate to newer models. The change follows GitHub's ongoing rotation of supported third-party models inside Copilot.
github.blog ↗