AI — Page 29

The state-approved pilot starts with two physicians reviewing every prescription and can move toward retrospective checks only after regulatory sign-off.

Olivia Moore's seventh consumer-app report adds U.S. card-spending data, with spending concentrated among people buying coding, creative and productivity software.

The MIT-licensed app coordinates coding agents, local models and visual checks, starting with Three.js browser games on macOS and Linux.

CEO Misha Laskin is pitching the model to companies and governments that want control over where AI runs; Reflection says Beam's weights will arrive later in October.

The framework OpenAI published on September 16th lets staff flag incidents and request public disclosure, while company teams control investigations and publication.

Founded by repeat entrepreneur Yoeven Khemlani and computer-vision researcher Harsha Khurdula, the model is available under Apache 2.0 and is designed to run on one 80GB GPU.

Policy chief Morgan Dwyer told New York City Council that OpenAI will publish its job-impact analysis; the federal bill would let companies submit workforce data to agencies.

Era generates linked company data across systems including Salesforce, Slack and Jira, giving developers a repeatable test environment without using production records.

Reflection AI's first scorecard puts Beam near or ahead of some open models on selected coding tests and behind others; its compute comparison excludes several costs of serving a model.

The Foundation found mostly sandbox edits and unsuccessful proxy attempts, and says it found no evidence its systems or data were compromised.

Tibo Sottiaux put the change at 50 tokens per second, up from 30, for subscribers using OpenAI products and supported partner apps.

After roughly six months in beta, Chakra is generally available. HackerRank says it ran more than 500,000 interviews, with Snowflake, Snorkel and Capgemini among its testers.

At a New York City Council hearing, an OpenAI speaker cited support for multiple bills but named neither the measures nor the company representative.

Creator Bin Liu is extending HyperFrames from an agent-written video framework into an editor where people can review and revise the same project.

BBC sources said Claude remained in Palantir's Maven Smart System as recently as last week, despite an August phaseout target.

The October 5th hearing brings representatives of OpenAI, Anthropic, Google and Meta before all 51 Council members, alongside whistleblowers including former Anthropic researcher Jacob Coxon. OpenAI sent the Council a letter last Thursday recommending safeguards ahead of the hearing.

Instinct users described recommendations tied to earlier conversations; the messages were not labeled as ads, and no evidence shows they generated commissions.

CMU safety researcher Ding Zhao, repeat founder Kyle Wong and engineer Simo Rachidi are building simulation tools for the risks robots encounter on real jobsites.

David Heinemeier Hansson says Ruby remains his choice for hand-written code; 37signals' October benchmarks show what its Rust port gains on a test machine.

A September 21st motion seeks records on Reddit's Compliance API, deletion events and whether content can be removed from models already trained on it.

GPT-6.1 Sol Pro leads the equal-weight index, which combines three task suites with samples ranging from 50 to 100 tasks.

Anthropic’s Claude Opus 5.5 scored 1.00 on the 50-task sample at $0.0242 per task. OpenAI’s GPT-6 Astra and GPT-6.1 Sol Pro tied for second at 0.99, with Sol Pro costing $0.0121 per task.

Reflection AI CEO Misha Laskin is preparing its first open-weight model for customer-controlled AI systems, backed by compute agreements with SpaceX and Nebius and a planned South Korean data center.

The open-source project routes local training through a central coordinator; its creator is testing a consumer-hardware approach, not claiming a finished chatbot.