Movements across frontier labs and models, and what they mean for security.
4 items from the 7-day rolling window: OpenAI bans accounts tied to a Russian influence operation, shares first results for its Jalapeño inference chip, and DeepMind looks back on 15 years of game AI research.
OpenAI disrupts a Russian influence campaign, ships its Jalapeño inference chip, and reaffirms zero data retention; DeepMind reflects on 15 years of game AI research.
This edition covers 12 items from OpenAI, Google DeepMind, and Google AI Blog, of which 6 were selected: OpenAI disrupts a Russian covert influence operation and ships a zero data retention option for frontier models, and DeepMind looks back on 15 years of games AI research.
OpenAI slows its model release pace for the "era of cyber-critical capabilities" and reaffirms zero data retention; DeepMind looks back on 15 years of games AI research and announces new game studio partnerships.
AI frontier developments over the past 7 days: new progress in DeepMind's games AI, OpenAI's AI Futures governance series, an update on zero data retention policy, and several pieces on cybersecurity and teen safeguards.
A roundup of AI frontier developments over the past 7 days: 19 items from OpenAI News, Google DeepMind Blog, and Google AI Blog, with actual publication concentrated between Aug 17 and Aug 21; focused on model safety pacing, zero data retention, and national security governance.
19 AI frontier items over the past 7 days, all from the ai_frontier category (OpenAI / Google DeepMind / Google AI Blog); the three ai_security sources (HiddenLayer / Lakera / Protect AI) still yielded nothing for lack of RSS. Highlights: OpenAI's new "AI Futures" blog series, pacing safety for frontier models, and a reaffirmation of the zero data retention policy.
16 selected vendor items from OpenAI and Google over the past 7 days. Security highlights: OpenAI reaffirms Zero Data Retention for the API and previews Private Safety Processing; publishes two articles on governing cyber capabilities, "Pacing model development in an era of cyber-critical capabilities" and "The Defender's Window"; Google DeepMind releases Gemini 3.7 Flash.
This edition covers 8 items published by OpenAI, Google DeepMind, and Google AI over the past 7 days, focusing on model release cadence and security-related statements; the three AI security sources HiddenLayer / Lakera / Protect AI again had no fetchable content this week (known gap).
This section is a 7-day rolling roundup (vendor blogs update weekly rather than daily). 16 items in total, ordered by publication time, most recent first.
The 7-day rolling window read 16 raw AI frontier items, of which 8 were selected; the three ai_security sources still have no RSS, so their permanent 0 rows in raw_items is a known condition.
This roundup covers the past 7-day window, though the material is concentrated in Aug 10–13 (nothing new from Aug 14–16; see the note at the end). Ordered by publication time, most recent first.
The 7-day window yielded 20 items in the `ai_frontier` category and 0 in `ai_security` (the three sources HiddenLayer / Lakera / Protect AI have no RSS and are not yet implemented in the fetch layer — a known gap, not evidence that there was nothing to report). 11 items selected, ordered most recent first, with items from the last 48 hours at the top; the remainder were enterprise case studies and marketing content, skipped as low-signal.
This edition is a 7-day rolling roundup (vendor blogs publish on a weekly cadence, so a 36-hour window is frequently empty). Material comes from OpenAI News and Google DeepMind/AI Blog; Anthropic News/Research had no new entries within this window.
Data sources: OpenAI News, Google DeepMind Blog, Google AI Blog (high-priority sources), 7-day window. The ai_security sources (HiddenLayer Research, Lakera Blog, Protect AI Blog) currently have no RSS collection; the corresponding entries in the database are zero this period — a known gap, not an anomaly.
This edition covers the past 7 days of updates from three high-priority sources — OpenAI, Google DeepMind, and Google AI Blog; Anthropic News/Research had no new content this period. Known gap: the three AI security sources (HiddenLayer Research, Lakera Blog, Protect AI Blog) have no RSS collection yet and contribute no content — a known condition, not an anomaly.
Sources: OpenAI News, Google AI Blog, Google DeepMind Blog (7-day rolling window — vendor blogs publish on a weekly cadence, so the 36-hour window is not used). Known gap: the three ai_security sources (HiddenLayer Research, Lakera Blog, Protect AI Blog) currently have no usable RSS and are not yet collected; their absence this period is expected, not an error.
AI vendor sources produced 5 records over the past 7 days; the most notable is OpenAI's preliminary evaluation of the Astra model's cyber capabilities. ai_security sources again yielded no data this period (a known collection gap).
Product and model updates from OpenAI / Anthropic / Google, with security-relevant items flagged (🛡).
Product and model updates from OpenAI / Anthropic / Google. 🛡 = security-relevant.
OpenAI / Anthropic / Google product and model updates · 🛡 = security-relevant
OpenAI / Anthropic / Google DeepMind product and model updates (🛡 = security-relevant)
OpenAI / Anthropic / Google product and model updates; 🛡 marks security-relevant items. Sunday (low volume); some items are cumulative developments from earlier this week.
Product and model developments at OpenAI / Anthropic / Google DeepMind (including security-related items 🛡)
Product and model developments at the three frontier labs; 🛡 marks security-related items.
Product and model updates from OpenAI / Anthropic / Google, with security-related items flagged (🛡).
Product and model developments at OpenAI / Anthropic / Google DeepMind · 🛡 = security-related
Product and model developments at OpenAI / Anthropic / Google, with particular attention to security. No major new releases in the last 24 hours; below are the ongoing July storylines.
🛡 = security-related
Product, model, and security developments across OpenAI / Anthropic / Google. 🛡 = security-related.