AI news · Thursday, October 8, 2026

OpenAI fires three safety researchers; claims they breached sensitive internal protocols

The internal culture at OpenAI is showing new signs of strain following the abrupt termination of three safety researchers, including Jasmine Wang, Tomek Korbak, and Mikita Balesni. The trio published an open letter disputing the company's claim that they mishandled confidential information, arguing instead that their work with external safety groups was both authorized and essential for monitorability. This incident has sparked fears among current staff that open dialogue on safety is being stifled, an irony not lost on observers given OpenAI’s history of championing transparency. Despite the friction, OpenAI remains central to broader industry debates: the Ethereum research community is currently discussing 'bunker mode' as a precautionary measure against AI models capable of solving complex cryptographic math problems that could theoretically compromise blockchain security.

Big Tech continues to navigate a turbulent relationship with the current administration, highlighted by a dramatic day for Microsoft. While Vice President JD Vance publicly condemned Microsoft for alleged abuses of the H-1B visa program, suspending the company’s ability to use the PERM pathway for green cards, CEO Satya Nadella appeared at the White House hours later to receive the National Medal of Technology and Innovation. This split—publicly punishing firms while privately celebrating them—is a recurring theme under the current administration, which also recently awarded National Medals of Science to other high-profile donors including Elon Musk, Jensen Huang, and Sergey Brin.

Hardware and agentic tools are also seeing a rapid shift in strategy. Amazon is officially killing its 'Fire' tablet brand, replacing the budget devices with a premium 'Alexa Tablet' lineup that runs full Android and features access to the Google Play Store. The new tablets prioritize Amazon’s 'Alexa+' AI assistant, which features 'On-Screen Intelligence' designed to analyze user screens to make contextual shopping recommendations. Industry analysts suggest this move acts as a strategic blueprint for a future Amazon-branded smartphone, positioning the company to compete more directly with Apple and Google in the agent-driven device space. Meanwhile, Google is pushing its own 'agentic' vision, launching an offline meeting-transcription tool called AI Edge Foresight that targets the crowded enterprise note-taking market, emphasizing that user data stays strictly on-device. Anthropic is simultaneously expanding its security footprint, launching an 'OSS Scanner' that uses its models to provide free vulnerability reports for open-source projects, though the company explicitly notes these reports come without human verification.

The quick hits

OpenAI fired three safety researchers for allegedly mishandling data — the employees claim they were punished for collaborating with outside experts, warning it will chill safety work company-wide.
The Trump administration suspended Microsoft and Adobe from a key visa program — a move that blocks the tech giants from sponsoring foreign workers for green cards while the government investigates claims of fraud.
Amazon is killing the Fire tablet brand to launch new Alexa-powered Android tablets — the move signifies a major pivot to premium hardware and Google Play support to better compete with the iPad.
Anthropic launched a free AI-powered security scanner for open-source projects — a move meant to help developers identify vulnerabilities at scale, though the model-generated reports currently lack any human review.

OpenAI fires three safety researchers; claims they breached sensitive internal protocols

The quick hits

Sources