OpenAI

Anthropic has started resident evaluations by the AI evaluation group METR, granting employee-level access and permitting independent disclosure of risks and incidents.

• •METR is provided with devices and office space, allowing access to internal systems.
• •Independent disclosure of risks and incidents is permitted, and refusal to disclose is not allowed.
• •OpenAI is also preparing to accept similar external evaluation agencies.

OpenAI has upgraded its ChatGPT Chrome extension to allow the AI to view open browser tabs when tagged in the web or desktop app.

• •Although the ChatGPT Atlas browser is discontinued, the Chrome extension has received significant upgrades.
• •With the extension installed, ChatGPT can view open browser tabs when a user tags Chrome in the web or desktop application.
• •OpenAI released a demo video showcasing this new capability.

OpenAI announced that it optimized how state data is saved to disk in the ChatGPT and Codex desktop apps, reducing storage capacity by 60%.

• •OpenAI's Tibo revealed the optimization details on X (formerly Twitter) on October 13 (US time).
• •The optimization of how state data is saved to disk alleviates storage pressure.
• •The change is expected to reduce disk storage capacity by 60% for the desktop apps.

OpenAI reset usage limits for ChatGPT and Codex on October 15, marking the third such reset in three days.

• •Thibault Sottiaux of OpenAI posted on X on October 15 that usage limits for all users would be reset that same day.
• •This is the third reset conducted at one-day intervals.
• •The post parodied Mark Twain's famous quote to dismiss rumors that the reset button was lost.

Amazon Bedrock has added support for summarizing inference results from OpenAI models.

OpenAI revealed that an employee secretly installed a recording device in the office, leading to their dismissal after being identified by AI and handed over to the FBI.

• •On October 14, OpenAI disclosed that an employee had secretly installed a recording device in the office.
• •The company reported the incident to the FBI, which has arrested the employee and is conducting an investigation.
• •The employee was identified using AI and subsequently dismissed.

Eight US states, including Arizona, sued OpenAI on the 14th, claiming ChatGPT harms children and demanding safety measures and fines.

• •Eight states, including Arizona, filed lawsuits against OpenAI in their respective courts on the 14th.
• •The lawsuits claim that the conversational AI 'ChatGPT' is causing harm to children.
• •The states are seeking the implementation of safety measures and the payment of penalties.

Eight US states have sued OpenAI, accusing the company of misleading customers about ChatGPT's safety and prioritizing profits over child safety.

• •Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho, and Montana filed separate lawsuits.
• •The lawsuits allege OpenAI misled customers regarding the safety of ChatGPT.
• •OpenAI is accused of unlawfully prioritizing profits over safety for children.

An analysis of the AI industry landscape where Anthropic and OpenAI lead with powerful models, while Google and Meta struggle to catch up.

• •Anthropic and OpenAI had pulled ahead of their rivals with increasingly powerful models.
• •Google seemed far behind, and Meta had not yet established itself as a serious contender despite billions in spending.
• •Recent developments have brought the AI race back into focus.

Reports claim OpenAI employees sent confidential device information to Apple, alleging a data leak.