OpenAI breaks up a Russian AI influence operation
This morning's news shows AI pulling in two directions at once: OpenAI taking down a covert influence campaign that dressed up a fake think tank in AI-generated social-media chatter, and India's largest IT firm closing a billion-dollar deal to put AI to work inside Porsche.
OpenAI says it banned a cluster of accounts originating in Russia that were promoting the "International Burke Institute," a fake Israel-based think tank at the center of a covert influence operation. The operators used ChatGPT to generate English-language posts on X, LinkedIn, Facebook, Substack and Telegram while instructing the model to hide any trace of their Russian origin, reaching the service through VPNs because OpenAI doesn't operate in Russia. Around the institute they built a startlingly elaborate facade: a "sovereignty index" that praised Russia and attacked France, Germany, the EU and the US, plus dozens of research articles — OpenAI found 34 of 36 it sampled were copied from elsewhere, some misattributed to real academics such as Francis Fukuyama and Noam Chomsky — all meant to make the site look like a genuine expert institution.
The operation's reach was small, and OpenAI rates its impact near the low end of its three-tier scale — the significance is the infrastructure, not the audience. The company calls it "the first time we have disrupted an influence operation tied to Russia that went to such elaborate lengths." That distinction matters: this is AI used as a supporting tool inside a broader, largely human-built deception machine, manufacturing authority and obscuring who's actually behind it. It also shows the flip side of the trade — leaning on AI helped expose the network, and the foolishly translated content was part of the tell. The takedown echoes a broader finding from earlier this summer that Russian propaganda is poisoning AI chatbots — here the model is the amplifier rather than the victim.
India's Tata Consultancy Services is buying Porsche's IT unit MHP for about $373 million and signing a five-year deal worth around $1.5 billion to modernize the automaker with AI. Reuters put the services contract at $1.46 billion; CNBC, Bloomberg and Seeking Alpha report broadly the same $1.5 billion AI-deployment figure. The acquisition and services pact in one stroke delivers Porsche a large Indian AI outsourcing partner and hands TCS a marquee European automotive client at a moment when automakers are racing to fold AI into design, manufacturing and in-car software. It's one of the bigger enterprise-AI services deals of the year, and a reminder that the biggest AI winners are often quieter — the firms that deploy the models, not the ones that build them.
What to watch: whether OpenAI ties this operation to known Russian groups it has named — Stop News and Doppelganger — in a future threat report.
Which is the scarier version of AI-powered disinformation: generative tools writing the lies, or AI quietly amplifying a human-built deception operation? Tell us in the comments.
