AI Frontier · Aug 28, 2026
7-day rolling window; this edition covers items published Aug 21–25. Vendor blogs don't publish daily, so this section rolls up on its own publication rhythm rather than as a daily snapshot. Known gap: the three ai_security sources (HiddenLayer Research, Lakera Blog, Protect AI Blog) have no RSS feed yet, so they contribute nothing here or in prior editions — that is expected, not a miss.
OpenAI bans ChatGPT accounts used in a Russian influence operation (Aug 25)
OpenAI banned a cluster of Russia-origin accounts that used VPNs to bypass access restrictions and generated ChatGPT content to promote the "International Burke Institute" (IBI), a fabricated think tank posing as Israel-based, along with narratives criticizing the West. Content was distributed across Substack, Telegram, X, Facebook, and LinkedIn.
Sources: OpenAI News
OpenAI publishes first results for its custom inference chip, Jalapeño (Aug 25)
OpenAI shared initial performance data for Jalapeño, its custom AI inference chip, claiming higher throughput, lower latency, and better power efficiency than existing options.
Sources: OpenAI News
OpenAI's CFO lays out the full stack behind "abundant intelligence" (Aug 25)
OpenAI CFO Sarah Friar explains how advances across chips, compute, models, and products compound to deliver more useful intelligence at greater scale and lower cost.
Sources: OpenAI News
Google DeepMind looks back at 15 years of game AI research, prototyping new gameplay with studios (Aug 21)
Google DeepMind published a retrospective spanning 15 years of AI research in games, from Atari to EVE Online, and described its ongoing partnerships with game studios to prototype breakthrough AI-driven gameplay.
Sources: Google DeepMind Blog