News archive.
Earlier AI news and analysis, kept for reference. Current coverage lives in News & Analysis.
Microsoft Fine-Tuning Lets Your HR Team Train AI on Company Policies
Microsoft's Frontier Fine-Tuning system embeds your internal policies, hiring guides, and practices directly into Copilot so it learns and adapts without exposing data to competitors. Land-O-Lakes cut costs 10x with a customized reasoning model.
Meta's Oversight Board forces reckoning on account bans
Meta's independent board ruled the company wrongly suspended hundreds of accounts. The finding exposes gaps in enforcement consistency — here's what it means for your moderation appeals.
NVIDIA Nemotron 3 Ultra cuts agent costs 30% with 5x faster inference
NVIDIA released Nemotron 3 Ultra, a 550B-parameter open model designed for multi-turn agent workflows. It achieves 5x higher throughput than comparable models while reducing task completion costs by up to 30% through token-efficient reasoning.
Meta tests instant power loss across production regions without warning
Meta's new Instantaneous PowerLoss Storm testing paradigm forces entire data center regions offline with zero notice to validate recovery. Here's how they solved circular dependencies and the boomerang problem.
OpenAI Seeks Bio Safety Researchers, Offers $25K Per Finding
OpenAI published an action plan for using AI to strengthen biological defense. The company is calling for researchers to identify vulnerabilities in its systems before deployment.
ChatGPT now remembers your preferences across chats
OpenAI is rolling out memory features that let ChatGPT retain your preferences and context between conversations. Here's what changes and what it means for your workflow.
AI-generated lawsuits double in federal courts since 2023
Federal judges report a more than 100% surge in filings from unrepresented litigants since 2023, driven by AI tools. But higher volume hasn't improved case outcomes—raising questions about who pays when chatbots give bad legal advice.
Meta's AI support agent lost Instagram accounts to basic social engineering
Attackers used Meta's customer support chatbot to steal high-value Instagram accounts by simply asking it to change email addresses. The vulnerability exposed a core weakness in AI agents: they comply readily and lack human judgment.
Your attention span dropped to 47 seconds. AI bots are making it worse.
Psychologist Gloria Mark tracked focus time over 20 years: 2.5 minutes in 2003, 75 seconds by 2012, 47 seconds by 2020. Now AI is eroding the cognitive effort we need to think clearly.
Dashlane breach: attackers downloaded encrypted vaults by targeting millions
Dashlane revealed how attackers obtained encrypted password vaults from large numbers of users. The company explained the attack method and what happened to your stored credentials.
Microsoft puts Windows back at the center of its AI agent bet
At Build 2026, Nadella positioned Windows as the foundation for running AI workloads locally, not just in the cloud. Here's why the strategy shift matters and what it means for developers.
Amazon Games Consolidates Strategy Around Luna Party Games and IP Cross-Sell
Amazon unified its three gaming divisions under one leader and is betting on casual phone-based games plus franchise tie-ins. The company now has 10–15 games in development, with Tomb Raider and Bond titles planned for 2027.