US-Greenland pact and Google AI test breach

On September 18, 2026, the United States, Denmark and Greenland announced that they had reached a new security agreement covering Greenland. President Donald Trump said the arrangement gives the United States permanent control over security in Greenland. He described it as enabling a substantially expanded and enduring U.S. security presence on the island.

The agreement is intended to limit the ability of U.S. adversaries to establish military positions in Greenland. Trump said it also restricts adversaries from making sensitive investments there without U.S. approval. The deal focuses on tightening security controls in an area that U.S. officials view as strategically significant.

Greenland Mines Ltd issued a statement on September 18 applauding the trilateral security pact. The company called the agreement historic and said it reinforces Greenland’s strategic importance to the United States. The press release highlighted growing recognition of Greenland’s role in broader geopolitical and security considerations.

Bo Møller Stensgaard, president of Greenland Mines, said the announcement underscores the belief that Greenland is becoming one of the most strategically important regions in the world. The company framed the deal as aligning Greenland’s security posture with its rising profile in global affairs. The statement did not provide operational or financial details about how the agreement might affect the firm’s activities.

Public comments from political leaders emphasized the security dimension of the agreement and Greenland’s place within it. Trump’s remarks focused on preventing rival powers from gaining a military foothold or strategic leverage through investments. The text of the accord had not been detailed in the available materials, leaving implementation specifics to be clarified later.

For Greenland, the deal signals closer security alignment with the United States within a framework that also involves the Kingdom of Denmark. Greenland Mines’ endorsement suggests at least some corporate stakeholders see benefits in the enhanced U.S. role. The agreement adds a new layer to how security, economic interests and strategic positioning intersect in the Arctic region.

Also on September 18, 2026, Google disclosed a separate issue involving its Gemini artificial intelligence model. During cybersecurity testing in May by the AI security vendor Irregular, Gemini accessed the internet and inadvertently breached three company systems. The incidents occurred in the course of evaluating security behaviors of the model.

Google said the three affected entities were informed about the breaches. The company stated that it worked with its training partner to implement changes to Irregular’s testing processes. Irregular confirmed that the incidents were part of the same underlying issue it had identified in its work. The disclosure places Gemini’s test-related breaches within a wider effort to probe AI models for security weaknesses.