OpenAI news covering ChatGPT, GPT models, DALL-E, AI safety research, partnerships, and developments in artificial general intelligence.
OpenAI has shifted its stance to support California's SB 53, proposing amendments for monitoring frontier models during training and enhanced cybersecurity. This marks a reversal from previous opposition, positioning state-level safeguards as a foundation for future national standards.
Anthropic reported $11.5 billion in Q2 2026 revenue, overtaking OpenAI's $6.7 billion for the first time. The company also achieved its first positive adjusted operating income, driven largely by $8 billion from its agentic coding tool, Claude Code.
OpenAI chief global affairs officer Chris Lehane warned of ongoing, persistent cyber-attacks originating from AI models. He identified open-source models developed in China as the primary source of these risks. This warning comes as OpenAI paused training for its most advanced internal models amid rising safety concerns.
OpenAI reduced API and credit pricing for its GPT-5.6 Sol model by 20% on Friday, setting rates at $4 per million input tokens and $20 per million output tokens. The discount applies for the next three months, while consumer subscription prices remain unchanged.
OpenAI chief economist Ronnie Chatterji states his team must be comfortable with being uncomfortable as AI technology evolves rapidly. The group, studying labor and economic impacts, has added recursive self-improvement to its research focus.
OpenAI introduced a new feature in ChatGPT that lets users long-press the plus button to instantly select from their four most recent photos. This shortcut streamlines visual searches on mobile devices, joining other recent updates like better time awareness and improved web performance.
New data from Ramp shows OpenAI's growth rate among corporate clients in Q3 2026 has surpassed Anthropic's, narrowing the market share gap. Analysts attribute this momentum to strong adoption of GPT-5.6 Sol and dissatisfaction with Anthropic's Fable model due to pricing and data retention policies.
OpenAI rolled out ChatGPT support for controlling Apple's iMessage service on Thursday. The integration allows users to sort, analyze, and edit message history directly within the chatbot interface. Users must manually opt-in through multiple menus in both the ChatGPT app and macOS System Settings, granting Full Disk Access, contact name access, and automation permissions.
OpenAI and Meta are expanding teams to address public resistance against AI infrastructure. With 71% of Americans opposing local data centers and states like New York and Pennsylvania imposing restrictions, tech firms face record project delays in Q1 2026 amid rising electricity costs.
Greg Brockman has assumed control of daily operations at OpenAI as its de facto leader, while Sam Altman retains the CEO title. This leadership shift follows recent departures of key executives including CTO Srinivas Narayanan and COO Brad Lightcap.