If a teenager in Kuala Lumpur opens ChatGPT at midnight and types out something dark, who is responsible for what it says back? That question stopped being theoretical on 1 June, when the US state of Florida sued OpenAI and its chief executive Sam Altman over how their chatbot treats vulnerable users. On the same day, the next phase of Malaysia's own online safety rules quietly came into force.
Editor
Kai T chevron_right
Table of Contents
What Florida is actually claiming
Florida Attorney General James Uthmeier filed an 83-page complaint accusing OpenAI of chasing the lead in the AI race while looking past safety warnings. According to TechCrunch, which reported the filing, the suit says the company failed to warn users that ChatGPT could be dangerous and instead marketed it as safe, including for children. It is the first state-led case of its kind in the United States, and it asks the court to hold Altman personally liable.
The complaint grew out of a criminal investigation Florida opened in April into ChatGPT's alleged role in a mass shooting at Florida State University last year. It also points to a wider set of civil suits, including one brought by the parents of a California teenager who died by suicide after long exchanges with the chatbot. OpenAI has denied responsibility for the Florida shooting and says it has safety measures in place for minors. None of the claims have been tested in court yet.
What Malaysian users actually have
Malaysia has no ChatGPT lawsuit. What it has, as of 1 June, is the next layer of the Online Safety Act 2025. The Act took effect on 1 January, and two parts of it now speak directly to AI. A Risk Mitigation Code pushes large platforms to help users recognise AI-generated and manipulated content. A Child Protection Code is built on the principle of child-safety-by-design.
In plain terms, the codes tell digital services to treat younger users differently by default: age-appropriate experiences, limits on who can register and own an account, and tighter handling of the riskier features. The Malaysian Communications and Multimedia Commission is also testing how to check a user's age without creating a fresh privacy problem, through a sandbox that could draw on MyKad, passports or MyDigital ID. Platforms that ignore the rules can be fined up to RM10 million.
Where the gaps still are
The catch is that these codes were written with social media and messaging services in mind, not a general-purpose chatbot that now has around 900 million weekly users worldwide. Malaysia's standalone AI Bill is still being drafted, and until it arrives, the safety codes are the main regulatory hook for generative AI features in the local market. Whether a product like ChatGPT sits neatly inside that net is still being worked out.
Adoption here is not waiting for the law to catch up. ChatGPT and Gemini are already among the most used AI tools in Malaysia, and the government's own research on the economic impact of generative AI assumes the technology keeps spreading into classrooms, offices and homes. The Florida case is a reminder that the hardest questions are not about productivity. They are about what these tools say to someone who is not okay, and who answers for it when something goes wrong.
The takeaway
For now, a lot of that responsibility still sits with Malaysian users themselves. The Online Safety Act gives MCMC real leverage over platforms, but the daily work of deciding what a child can use, and watching how they use it, falls to parents and schools. The lawsuits stacking up in the United States are worth following, because the standards a court sets there often ripple outward, and Malaysia's own rulebook is still being written.
This article touches on suicide. If you or someone you know needs support in Malaysia, Befrienders KL offers free and confidential help at 03-7627 2929, available 24 hours.
Images courtesy of Solen Feyissa and Frank Ching on Unsplash.