TECHNOLOGY · AI · CYBERSECURITY · THE SYSTEMS SHAPING BUSINESSPage 2 of 3: earlier stories.
The Allen Institute for AI's open-source training framework targets trillion-parameter mixture-of-experts models and posts a throughput lead over Nvidia's Megatron-core in Ai2's own test.
Search results are billed at the partners' list prices from existing AI Gateway credits, and providers must follow Cloudflare's verified-bot and robots.txt rules.
The cheaper desktop AI system goes on sale Oct. 23 through six PC makers, and two linked units can run models of about 200 billion parameters.
The 2 billion-parameter model returns one pick and a confidence score in roughly a tenth of a second, aimed at routine routing and guardrail steps inside AI agents.
Some Google employees reportedly see a gap between the model's benchmark scores and its everyday work, a day after Gemini 4 Argon reached its first users.
The public preview reaches the Copilot CLI and desktop app on macOS and Windows, with per-app approvals and an off switch for organizations.
The British bank is extending Claude from a staff knowledge tool and email triage into software engineering and legacy modernization, with a majority of engineers expected on it in 2027.
Chris Degnan, Snowflake’s longtime revenue chief, denies sharing secrets and says he resigned before joining the rival AI coding startup as its CRO.
Leaked npm tokens were almost all dead, but more than half of exposed Google Cloud service account keys still worked, the researchers found.
RSS goes dark on Nov. 13 and the public API in March 2027, pushing researchers and AI developers toward registered apps or paid data deals.
Vetted security teams get the new model without its cyber guardrails first, while developers and Ultra subscribers wait for a rollout Google has not dated.
Attackers used a fresh JFrog Artifactory flaw three days after disclosure, and the break-in went unnoticed for about two weeks.
The Chinese AI lab is releasing a high-level language port plus compute and communication libraries for Huawei's Ascend 950, a direct challenge to CUDA's grip.
A default trust_remote_code setting turned a routine metadata check into code execution; the fix shipped in June, but no CVE or advisory was issued.
Born inside OpenAI and now run by a foundation with Red Hat and Nvidia on board, the MIT-licensed project targets companies moving from chatbots to always-on agents.
Developers can now build and run Linux containers through WSL without third-party software, while IT teams get Defender visibility and Intune registry controls.
A companion Sign in with ChatGPT feature lets Plus and Pro subscribers use their plan allowance inside 16 outside tools, but OpenAI has not set up app-store style billing.
The DevDay release keeps GPT-6 Sol's token rates, halves the cached-input price and pairs it with a 300-token-per-second option billed at six times standard rates.
The DevDay update moves OpenAI's coding agent further toward long-running cloud work and folds its Daybreak Blue security models into repository scanning.
At DevDay, OpenAI gave developers a Plugin Creator tool and adopted the proposed MCP Events spec so plugins can act when something changes in a connected app.
Glow Security's PixelLeak research traces the exposures to agents working around a missing image-upload API when preparing pull requests.
Power BI Pro and Premium Per User customers get hosted app building and a Fabric database at no extra cost, a bid to pull millions of analysts onto Fabric.
An NBER paper backs out the figure from how share prices moved for companies with large engineering payrolls, and translates it into a 3.61% lift to GDP.
Exposed tables held valet customers' license plates, plaintext passwords and private messages, as AI-built apps ship without row-level security.