AI moves into compliance and cyber

Companies are starting to use generative artificial intelligence for tariff math and trade paperwork, while OpenAI is rolling out a cybersecurity model to vetted defenders. (businessinsider.com) (openai.com) Business Insider reported on April 15 that United States importers are using artificial intelligence tools from firms including KPMG, EQI and Gaia Dynamics to model tariff scenarios, identify possible overpayments and reduce compliance errors after the Supreme Court’s February 2026 tariff ruling. (businessinsider.com) (cnbc.com) That ruling, in a 6-3 decision, said President Donald Trump lacked authority under the International Emergency Economic Powers Act to impose his sweeping April 2025 tariffs, setting up refund claims and new accounting questions for importers. (cnbc.com) (deloitte.com) Trade compliance is the work of matching products, countries, duty codes and legal rules to each shipment. Companies told Business Insider that generative artificial intelligence is being used to scan those records faster and test “what if” tariff outcomes across suppliers and routes. (businessinsider.com) OpenAI made a parallel move on April 14, saying it was expanding its Trusted Access for Cyber program to thousands of verified individual defenders and hundreds of teams that protect critical software. (openai.com) (helpnetsecurity.com) The company also introduced GPT-5.4-Cyber, a version of GPT-5.4 fine-tuned for defensive cybersecurity work, including tasks such as binary reverse engineering and vulnerability analysis for approved users. (openai.com) (cyberscoop.com) Cybersecurity models are useful because defenders often need help reading malicious code, tracing how an attack works and checking software for weaknesses before criminals find them. OpenAI said the new model is “cyber-permissive,” meaning it is tuned to be more helpful on legitimate security tasks while still gated behind verification and safeguards. (openai.com) (helpnetsecurity.com) The rollout also lands in the middle of a competition over who gets access to the strongest security models. CyberScoop said OpenAI’s move puts it in direct competition with Anthropic’s restricted cyber offering and raises questions about how widely advanced offensive-adjacent tools should be distributed. (cyberscoop.com) (securityweek.com) On the trade side, the immediate pressure is less about hacking than paperwork: companies are trying to recover tariff payments, update contracts and plan for fresh duties imposed under other legal authorities after the court decision. (thomsonreuters.com) (deloitte.com) Taken together, the two moves show where generative artificial intelligence is being deployed first: not as a general assistant, but as a tool for specialists working through dense rulebooks, high-volume records and high-stakes decisions. (businessinsider.com) (openai.com)