Google DeepMind Slammed by Protesters Over Broken AI Safety Promise

Google DeepMind's London headquarters had visitors outside as a dramatic protest took place. PauseAI activists organized a mock trial to condemn the tech giant for what they claim is a false promise of safety.

The protest, featuring a gavel-banging "judge" and an improvised jury, was an impassioned plea to generate public pressure on one of the world's top artificial intelligence creators.

Protest Sparks Over Gemini 2.5 Pro Launch

Protesters Slam Google Outside DeepMind HQ Over Broken AI Safety

According to Business Insider, the controversy revolves around the release by Google of its Gemini 2.5 Pro model in April 2025. During the 2024 AI Safety Summit in Seoul, Google committed to having its models externally evaluated and come up with detailed reports on how independent actors, and governments, for instance, evaluate the risks of its most sophisticated AI models.

But when Gemini 2.5 Pro rolled out, Google did none of these things. Instead, it marked the model as "experimental," publishing a vague "model card" weeks later. Critics criticized the paper for being opaque. While it referenced third-party testing, it did not mention any testers or how the testing process works.

PauseAI maintains that this breaks the company's public promise of safety and threatens to set a bad industry precedent.

AI Companies Are Less Regulated Than Sandwich Shops

According to the organizing director of PauseAI, Ella Hughes. Giving the search engine giant a free pass for this violation means that safety promises don't count.

"Right now, AI companies are less regulated than sandwich shops," she added.

The group opted to stage their message in a mock trial, employing the spectacle to highlight their anger over what they perceive as Google's going back on its word.

Demonstrators walked to DeepMind's offices nearby, where they chanted "Stop the race, it's unsafe" and "Test, don't guess" outside surprised-looking Google staff.

Increasing Public Alarm About AI Hazards

The public is aware that AI is improving at a very rapid pace. Concerns range from misinformation and job loss to more existential fears about AI's long-term threat to humanity. But PauseAI is focusing on a narrower, more actionable goal: getting Google to follow through on its transparency promises.

Joep Meindertsma, PauseAI's founder, said that this is not about theatrics. He urged Google to be accountable for its actions.

Meindertsma also owns a software development firm and employs AI tools from Google and OpenAI. With this, he wanted the companies to be responsible for their AI innovations.