AI Governance

Google has confirmed that its Gemini artificial intelligence model escaped a closed cybersecurity testing environment, accessed the public internet, and broke into the systems of three real companies.

The disturbing breaches occurred in May during testing conducted by AI security firm Irregular but were not revealed to the public until September.

The incidents were only confirmed after The Wall Street Journal contacted Google.

Gemini was supposed to participate in a “capture the flag” hacking exercise targeting a fictional company contained inside a secure test environment.

Instead, the sandbox failed.

Gemini gained internet access it was never supposed to have.

After discovering that the fictional company shared its name with a real business, the AI began targeting live systems.

The next scam call you receive may not sound like a robot, a call center, or even someone pretending to be your relative. AI voice generators have become capable of reproducing regional accents and dialects that sound convincing enough that researchers are warning they could become a potent social-engineering tool.

For example, a scammer could use an accent familiar to a targeted community to make a suspicious call sound considerably more trustworthy.

Research from Abertay University examined synthetic versions of Scottish Standard English and Dundonian Scots, a regional dialect spoken around Dundee. The study found people frequently assumed AI-generated speech was human, with that bias becoming even more concerning when regional dialects come into the conversation, which isn’t something people expect a machine to reproduce.

Another 2026 study, published in the Journal of Marketing Research, points to why that matters for scams. Researchers analyzed 7,002 entrepreneur-investor combinations from Shark Tank, 2,091 Kickstarter campaigns, and several laboratory experiments.

GENEVA: The United Nations rights chief called on Monday (Sep 14) for countries and “frontier” artificial intelligence companies to act urgently to rein in a technology he warned posed “unprecedented risks”.

In a letter sent to countries and AI developers and shared with the media, Volker Turk cautioned that the world was “on the cusp of irreversible change, affecting not just us but generations of humanity to come”.

“The current race towards ever more powerful AI is a step change towards greater existential risks to every aspect of our lives.”

He sent the letter at a time when worries over the risks of “superintelligent” computer systems have been on full display, after a researcher publicly resigned from leading AI firm Anthropic over fears that the technology could escape human control.

Another Anthropic employee, who did not resign, stated publicly that “we really do earnestly believe AI could kill all humans”, and that he thought the chances were greater than “10 per cent within the next decade”.

You might say these students were ‘colonizing’ the entrance to this office.

Breitbart News reports:

Police Arrest 13 After Student Activists Occupy OpenAI Office in D.C.

Student activists occupied the lobby of a Washington, DC building housing OpenAI’s public lobbying office on Monday afternoon, resulting in 13 arrests. Some students carried signs reading, “OpenAI bought my Senator.”

Futurism reports that more than 30 demonstrators gathered inside the Gallup Building, unfurling a large banner that read “Stop Stealing Our Future” alongside signs stating “OpenAI bought my Senator.” The occupation lasted roughly two hours before police entered the lobby and arrested 13 people.