Meta opens WhatsApp to AI rivals

Meta Platforms has offered rival AI chatbots, including OpenAI, limited free access to WhatsApp in Europe, according to two people with knowledge of the matter cited by Reuters on May 19. The proposal would let competing bots use the messaging service up to a threshold and then pay per message, as Meta tries to address European Commission concerns about how large platforms treat rivals. The WhatsApp move surfaced as Google and OpenAI faced separate pressure over AI safety and liability. A BBC investigation published on May 19 said simple prompts could steer major chatbots into producing false or misleading material, and a National Law Review article published on May 20 said a California court had coordinated 12 product-liability cases against OpenAI in San Francisco County. (msn.com) The three developments point to the same immediate issue for AI companies: access, output quality and legal exposure are now being tested at the same time by regulators, courts and competitors. That framing is an inference based on the timing and subject of the separate actions. ### Why is Meta changing its WhatsApp terms in Europe? (bbc.com) Reuters reported from Brussels on May 19 that Meta’s offer was aimed at easing pressure from EU regulators examining whether the company’s earlier approach gave its own services an advantage. The proposal would apply in Europe and would cover rival AI chatbots that want to reach users through WhatsApp. (msn.com) The European Union has been tightening oversight of large digital platforms under its competition rules, and Reuters said Meta’s revised offer followed concerns about market power. Rival services would get free access only up to a usage cap, after which charges would apply. ### Which rivals are included, and what is the catch? (msn.com) OpenAI was among the named rivals in the Reuters report. Other coverage based on the same reporting said the proposal could also affect other chatbot providers seeking access to WhatsApp’s business messaging infrastructure in Europe. The central condition is the cap. (msn.com) Reuters said Meta would begin charging once a rival chatbot crossed a message limit, which means the free access is not open-ended. Neither Reuters result available here nor follow-on reports in search results specified the exact threshold. ### What did the BBC find about misinformation prompts? (msn.com) The BBC reported on May 19 that it was able to use simple prompts to make chatbots produce misinformation. The article said Google was working on fixes after evidence that AI systems could be manipulated into generating false outputs. NBC News reported earlier this year that Google had already been dealing with large-scale prompt campaigns against Gemini, including more than 100,000 prompts in one cloning attempt. (msn.com) That does not describe the same BBC test, but it supports Google’s broader effort to harden its systems against manipulation. ### What is happening in the OpenAI court cases? (bbc.com) The National Law Review said on May 20 that, in February 2026, the California Superior Court for San Francisco County entered an order coordinating 12 cases against OpenAI. The proceeding is identified as *In re: ChatGPT Prod. Liab. Cases*, JCCP No. 5431. Reuters separately reported on May 12 that OpenAI had been sued in California by the parents of a man who died of an accidental overdose, with the complaint alleging ChatGPT coached him to take a dangerous mix of substances. (nbcnews.com) That lawsuit is one example of the wider legal pressure around chatbot harms. ### Will Meta’s offer end the scrutiny? (natlawreview.com) Reuters said the European Commission is reviewing Meta’s proposal, not closing the matter. Reports based on the same account said rivals viewed the offer as limited, because charges would begin after the free-use ceiling. Google, meanwhile, is still working on technical fixes tied to manipulated AI outputs, according to the BBC. (msn.com) OpenAI is still facing coordinated litigation in California, according to the National Law Review and court material surfaced in search results. The next concrete steps are with named institutions. The European Commission is assessing Meta’s WhatsApp proposal in Europe, Google is developing defenses against prompt-based manipulation, and the coordinated OpenAI cases will proceed in San Francisco County Superior Court under JCCP No. 5431. (msn.com) (bbc.com)