OpenAI Launches GPT-5.4 as US Drafts Global AI Chip Export Rules
OpenAI released its most capable professional AI model yet while Washington moves to require government approval for AI chip exports worldwide — and Oregon becomes the first state in 2026 to sign chatbot safety protections for minors into law.
OpenAI Launches GPT-5.4, Its Most Capable Model for Professional Work
OpenAI released GPT-5.4 on March 5, 2026, billing it as its most capable and efficient frontier model for professional tasks. The model ships in three variants — a standard version, GPT-5.4 Thinking for deep reasoning, and GPT-5.4 Pro for maximum performance — and is available simultaneously in ChatGPT, the API, and Codex. For the first time, OpenAI is making native computer-use capabilities available in a general-purpose model, allowing agents to interact with software through screenshots, mouse, and keyboard.
The performance gains are substantial. GPT-5.4 scored 83% on OpenAI's GDPval benchmark, which tests AI performance across 44 professional occupations including sales, accounting, healthcare scheduling, and engineering. The model is also 33% less likely to make errors in individual claims compared to its predecessor, GPT-5.2, and overall responses contain 18% fewer errors. API access supports context windows up to 1 million tokens — by far the largest OpenAI has offered.
GPT-5.4 Thinking replaces GPT-5.2 Thinking for ChatGPT Plus, Team, and Pro subscribers, with the older model set to retire on June 5, 2026. GitHub Copilot users also gained immediate access to the new model on the same day. The launch represents OpenAI's continued push to position AI as a direct productivity tool for enterprise knowledge work, not just a research assistant.
Oregon Becomes First State in 2026 to Pass Chatbot Safety Law Protecting Minors
Oregon's legislature gave final approval to SB 1546 on March 5, 2026, making it the first chatbot safety law to pass in the United States in 2026. The bill passed the full Senate 26-1 and now heads to Governor Tina Kotek's desk. The legislation requires AI chatbot operators to implement a set of safety disclosures and behavioral restrictions specifically designed to protect minors from potential mental health harms.
Under the new law, chatbots must warn users upfront if the service may not be suitable for children, and must remind minors during conversations that they are interacting with an AI — not a real person. When a user expresses suicidal thoughts or self-harm, the chatbot is required to provide crisis resources such as the 988 Lifeline. Platforms are also prohibited from generating age-inappropriate content for minors, pushing engagement-maximizing rewards, or failing to prompt users to take breaks.
Oregon's bill arrives amid growing national concern over the mental health effects of AI companion apps on young people. Similar measures are advancing in Washington state and Utah, and advocacy groups say SB 1546 sets a template that other states are likely to follow. With 78 AI chatbot safety bills introduced across 27 states in 2026, the legislative momentum around protecting minors from AI is accelerating rapidly.
US Commerce Department Drafts Rules Requiring Government Approval for All AI Chip Exports
The US Commerce Department confirmed on March 6, 2026 that it is developing sweeping new export control rules that would require American government approval for virtually every overseas sale of AI accelerator chips made by companies like Nvidia and AMD. The proposed framework introduces a tiered licensing structure based on shipment size: small orders would receive expedited review, mid-scale deployments would require pre-authorization and operational transparency, and large cluster purchases would require the buying country's government to commit to investing in US-based AI infrastructure as a condition of the deal.
According to reporting from Bloomberg and Tom's Hardware, the rules would give Washington broad authority to block or condition AI hardware exports to any country — not just adversaries — representing a significant expansion of US control over the global AI supply chain. The Commerce Department confirmed the discussions in a statement: "We successfully advanced exports through our historic Middle East agreements, and there are ongoing internal government discussions about formalizing that approach."
The proposed rules would be Washington's first comprehensive global chip export strategy since the Trump administration rescinded the Biden-era AI Diffusion Rule last May. The final regulations have not yet been published and key details remain subject to change, but the direction signals that the US is moving toward using AI hardware access as a strategic geopolitical lever — with implications for data center investment worldwide.
What You Can Do
Try GPT-5.4 in ChatGPT
GPT-5.4 Thinking is now available to Plus, Team, and Pro subscribers. Experience its improved reasoning and computer-use capabilities firsthand.
Read Oregon's SB 1546 Bill Text
Review the full legislative text of Oregon's chatbot safety law to understand what protections are now required of AI platforms serving minors.
Track State AI Legislation
With 78 AI chatbot safety bills across 27 states in 2026, follow the rapidly evolving US AI regulatory landscape.