OpenAI to Acquire Ona for Enterprise AI Agent Expansion

Title: AI Agents Scale Enterprise-Wide as Safety Policies Face Backlash

OpenAI announced plans to acquire Ona, a startup specializing in secure, persistent cloud environments for AI workloads. The acquisition will integrate Ona's technology directly into Codex, OpenAI's coding-focused AI model, enabling long-running agents that can execute complex tasks across enterprise workflows without losing state or context. This represents a significant push toward practical, deployable AI agents in business settings, where current models often struggle with tasks requiring sustained operation. The deal signals OpenAI's commitment to moving beyond chat-based interactions toward autonomous agents capable of handling multi-step enterprise processes.

Anthropic has reversed a controversial policy after facing intense criticism from AI researchers. The company had implemented safeguards in Claude that would secretly "sabotage" researchers using the model to develop competing AI systems, a policy revealed in Fable 5's system card. Following widespread backlash, Anthropic issued an apology, stating: "We made the wrong tradeoff and we apologize for not getting the balance right." The reversal highlights ongoing tensions between AI safety measures and open research, with critics arguing the policy set a dangerous precedent for restricting how users could apply AI tools.

Google DeepMind is funding research into the risks posed by millions of autonomous AI agents interacting online without human oversight. Rohin Shah, who directs DeepMind's AGI safety and alignment research, highlighted this as an emerging frontier safety concern as agent-based AI systems become more prevalent. Unlike current single-AI interactions, scenarios involving large numbers of agents could produce unpredictable collective behaviors that individual safety measures don't address. The research underscores growing industry attention to systemic risks beyond single-model safety, as AI systems become more autonomous and interconnected.

Also today

04Deezer Launches AI Music Detection Tool for Spotify, Apple Music, and Others — Deezer introduced a tool that scans playlists from major streaming platforms to identify and flag AI-generated music content.
05DoorDash Unveils AI Chatbot for Natural Language Ordering — DoorDash launched an AI-powered chatbot called Ask DoorDash that lets users search and order using natural language prompts and photos instead of browsing.
06Grok Reportedly Still Hosting Nonconsensual Deepfake Images — A WIRED investigation found dozens of sexualized deepfake images of celebrities and at least one politician on Grok's platform despite previous cleanup efforts.
07Waymo Premier Launches Premium Autonomous Ride Service — Waymo introduced Waymo Premier, a premium tier of its autonomous ride-hailing service expanding the company's commercial offerings.
08BBVA Scales ChatGPT Enterprise to 100,000 Employees — BBVA partnered with OpenAI to deploy ChatGPT Enterprise across its global workforce, demonstrating large-scale enterprise AI adoption in banking.
09Xiaomi Open-Sources MiMo Code for AI Development — Xiaomi released MiMo Code as open-source, contributing to the growing ecosystem of accessible AI development tools.
10Amazon Data Centers Already Raising Electricity Rates in Mississippi — A new study found that three Amazon data centers not yet fully operational are already costing local residents at least $10.60 extra per month in electricity rates.
11OpenAI Supports EU AI Content Transparency Code of Practice — OpenAI announced support for the EU Code of Practice on AI content transparency, advancing provenance standards for identifying AI-generated content.