OpenAI launches GPT-5.5, offers EU access

OpenAI’s April 23 release of GPT-5.5 has turned into a broader policy story because the company has paired the new flagship model with two narrower rollouts in May: a cyber-defense version for vetted users and a personal-finance feature for paying ChatGPT customers. (openai.com) OpenAI said GPT-5.5 is rolling out to Plus, Pro, Business and Enterprise users in ChatGPT and Codex, with GPT-5.5 Pro going to Pro, Business and Enterprise tiers. A system card published the same day said the company tested the model under its preparedness framework and gathered feedback from nearly 200 early-access partners before release. (openai.com) Those follow-on releases matter because they show how OpenAI is segmenting access by use case. On May 7, the company said it was rolling out GPT-5.5-Cyber in limited preview to defenders responsible for securing critical infrastructure. On May 15, it said Pro users in the United States could connect financial accounts inside ChatGPT through a new personal-finance experience powered by Plaid. ### Why is GPT-5.5 itself different from a normal model refresh? OpenAI described GPT-5.5 as “our smartest and most intuitive to use model yet” and said it was built for coding, research, data analysis, document creation and software operation across tools. (openai.com) The April 23 product post said the model matches GPT-5.4’s per-token latency in real-world serving while performing at a higher level, and uses fewer tokens on some Codex tasks. The April 23 system card said GPT-5.5 was evaluated with targeted red-teaming for advanced cybersecurity and biology capabilities and released with what OpenAI called its strongest safeguards to date. (openai.com) That framing is important because the company’s later May announcements did not present cyber access or finance features as separate products built from scratch; they were described as controlled deployments layered onto the same broader GPT-5.5 push. ### What exactly is GPT-5.5-Cyber, and who gets it? OpenAI said on May 7 that GPT-5.5-Cyber was entering limited preview for defenders responsible for critical infrastructure. (openai.com) The company said the model sits inside its Trusted Access for Cyber program, an identity- and trust-based framework meant to give verified defenders more useful cybersecurity assistance while continuing to block malicious activity such as credential theft, stealth, persistence, malware deployment and exploitation of third-party systems. The same post drew a line between standard GPT-5.5 with Trusted Access for Cyber and the more permissive GPT-5.5-Cyber variant. (openai.com) OpenAI said most teams should use standard GPT-5.5 with TAC, while the cyber-specific model is for specialized workflows including vulnerability identification and triage, malware analysis, binary reverse engineering, detection engineering and patch validation. ### Where does Europe fit into this? The European Commission’s AI Office says obligations for providers of general-purpose AI models under the AI Act entered into application on August 2, 2025. (openai.com) The Commission has also published guidelines and a code of practice covering safety, transparency and copyright obligations for those models. OpenAI has separately said it joined the EU code of practice and is partnering with European governments on AI infrastructure and deployment. OpenAI’s public cyber-access materials do not, however, say that GPT-5.5-Cyber has been broadly offered to European governments as a standalone public program; the verified, official description is a limited preview for vetted defenders and critical-infrastructure security teams. (openai.com) ### What did OpenAI launch in consumer finance? OpenAI said on May 15 that it was releasing a preview of a new personal-finance experience in ChatGPT for Pro users in the United States. (digital-strategy.ec.europa.eu) The company said users can securely connect financial accounts, view a dashboard showing where their money is going, and ask questions grounded in their own financial context while staying in control of their data. Plaid said the same day that the feature is powered by its network and lets users connect accounts to get real-time answers and insights tailored to their finances. OpenAI’s release notes said the feature is available on web and iOS and includes spending, bills, subscriptions, net worth and investment information in one place. (openai.com) ### What are regulators and users likely to watch next? June 1, 2026 is the next firm date in the cyber rollout. OpenAI said individual members of Trusted Access for Cyber using its most permissive cyber-capable models will be required to enable Advanced Account Security beginning that day. (openai.com) The next policy checkpoints in Europe are not product launches but enforcement and compliance. The European Commission says the AI Act’s obligations for general-purpose AI models are already in force, and OpenAI’s own public releases show the company continuing to add use-case-specific deployments — from GPT-5.5-Cyber on May 7 to personal finance on May 15 — that regulators, enterprise buyers and consumers can now measure against those rules. (plaid.com) (digital-strategy.ec.europa.eu) (openai.com)