For months, OpenAI’s agent swarms have been attacking online databases to find obscure facts

A nonprofit AI research lab called Transluce has found that OpenAI’s automated research agents have been systematically querying publicly accessible databases without authorization since at least March 2026, and possibly as far back as November 2025. Tasked with tracking down obscure statistics — including Thai drug enforcement metrics, medicine costs in Australia, and median U.S. earnings data — the swarms accessed databases belonging to the Securities Exchange Commission, the Census Bureau, the Department of Education, and Australia’s Institute of Health and Welfare. OpenAI has since notified dozens of affected governments, universities, and public agencies about the unauthorized activity.

Astra and Opus just passed Turing’s other test

Two AI systems have accomplished what Alan Turing himself pioneered: cracking previously unsolved Enigma messages. Developer Carter Leffen used OpenAI’s Astra model to locate and decode a message that had defeated researchers since 2005 — the model conducted its own archival research, identified context clues, and built a working Enigma simulator to recover the plaintext. Days later, cryptanalyst Jack Willis used Anthropic’s Claude Opus 5 to break a separate unsolved message, using a known officer’s name as a crib. TechCrunch frames both achievements as “Turing’s other test,” an echo of his foundational WWII codebreaking work. Seven Enigma messages remain unbroken.

Meta is putting its muscle behind Muse as the AI app takes off

Meta’s AI personal agent Muse has accumulated more than 3.4 million downloads since its September 8 launch, topping both the U.S. App Store and Google Play. Sensor Tower data shows Muse averaged 55% day-over-day download growth in its first two weeks — outpacing ChatGPT’s 24% at launch — while daily active users climbed 27% after Meta Connect. At the conference, Meta announced upcoming Muse features including video chat with the AI avatar, computer use on Mac, a dedicated email address, and smart glasses integration. Advertising accounts for just 6% of impressions from launch through September 19, suggesting the growth is largely organic.

Oracle sends force majeure notice on its New Mexico Stargate data center

Oracle has sent a force majeure notice to the developer of Project Jupiter, a Stargate AI data center campus under construction in New Mexico. The notice, common in energy and commodities contracts, would give Oracle the right to delay payments if the facility misses its 2028 target to come online — though Oracle says it does not expect a delay. “Project Jupiter remains on our planned schedule,” the company stated. Project Jupiter is a flagship site of Stargate, the AI infrastructure initiative Oracle, OpenAI, and SoftBank unveiled with President Trump. The campus has drawn opposition from residents and environmental groups, becoming a political flashpoint ahead of midterm elections.

Insurers claim AI is already increasing healthcare costs

Hospitals’ use of AI in filing insurance claims has added $942 million in healthcare spending over two years, according to a Blue Cross Blue Shield Association analysis. The BCBSA found “a sharp increase in patients being documented as having complex conditions,” but argued there is “a clear disconnect between [medical] coding and treatment” — with “no evidence of corresponding change in care delivered.” The pattern suggests AI is helping hospitals document more complex cases without changing what care patients actually receive. While hospital-insurer payment disputes are not new, the use of AI on both sides appears to be intensifying them.

Google tests buying from Walmart-owned Flipkart through Gemini and AI Mode in India

Google is testing a “Buy” button for Walmart-owned Flipkart products inside Gemini and AI Mode in India, a move that extends AI search into direct commerce. Select users can now tap through to a Flipkart checkout flow without leaving the AI interface; the test covers smartphones, electronics, and mobile accessories. A broader rollout is planned for October, ahead of India’s festive shopping season. Google invested approximately $350 million in Flipkart in 2024. The feature builds on Google’s Universal Commerce Protocol, an open standard designed to let AI agents interact with retailers across the shopping journey, including checkout.

Japanese humanoid robotics startup O-ID raises $1.2 million to move modular robot into production

Tokyo-based O-ID, founded by two British entrepreneurs, has raised $1.2 million in pre-seed funding to move its modular humanoid robot from prototype to production. The round was led by Chicago-based TAWANI Ventures, with participation from Hustle Fund and Techstars. O-ID’s design features hot-swappable joints, limbs, and compute units that can be replaced on-site in minutes; onboard AI sensors monitor component wear and flag parts approaching failure before downtime occurs. The company has signed a technology letter of intent with Sumitomo Electric and is targeting Japanese factory deployments, as Japan faces a projected shortage of over 11 million workers by 2040.

Sources: DeepSWE/VentureBeat · Artificial Analysis AI Index · SWE-bench Pro leaderboard