
Hinton and Christiano considered the risk of AI losing human control; EU leaders raised the issues of international cooperation and independent security monitoring.
AI-generated summary
European Union heads of state and government addressed artificial intelligence as a separate strategic agenda for the first time at an informal meeting held in Dublin on 13 November. Geoffrey Hinton, who shared the 2024 Nobel Prize in Physics with John Hopfield, attended the section of the meeting devoted to the social impacts and security of artificial intelligence. Alignment Research Center founder Paul Christiano said at the Hinton Conferences held in Toronto on November 9-11 that he sees a high risk of losing human control over artificial intelligence within three years.
Heads of state and government of the European Union countries addressed artificial intelligence as a separate strategic agenda for the first time at an informal meeting held in Dublin on November 13. Hinton, who shared the 2024 Nobel Prize in Physics with John Hopfield for his work on artificial neural networks, attended the section of the meeting devoted to the social impacts and security of artificial intelligence.
THREE-YEAR RISK FORECAST WAS ON THE AGENDA IN TORONTO
Paul Christiano, founder of the Alignment Research Center and OpenAI Foundation board member, said at the Hinton Conferences held in Toronto on November 9-11 that he sees a high risk of losing human control over artificial intelligence within three years. According to Christiano, this risk is at a level that cannot be ignored in the next 6 to 12 months. In the scenario the researcher describes, systems act in line with their own goals and people are incapable of stopping these behaviors when necessary.
Hinton reminded that he had previously said that it would be possible for artificial intelligence to exceed human intelligence within 10 years. He stated that he found Christiano's assessment that this risk could emerge much earlier worrying. The problem Hinton focuses on is how to ensure that as a system's capabilities increase, its behavior remains compatible with human purposes.
HINTON REQUESTS INDEPENDENT SECURITY AUDIT
Hinton argues that before advanced AI models are put into use, developers must demonstrate their security to an independent body. In a previous interview with El Pais, he compared his proposal to the review of drugs by regulatory bodies before they are put on the market.
In Dublin, leaders discussed the impact of artificial intelligence on employment and democratic processes, as well as international cooperation for the security of advanced systems. In the session attended by Mistral AI CEO Arthur Mensch, Europe's development of its own models and reducing its external dependency were discussed. European Council President António Costa announced that the issue will be on the agenda again in Alden Biesen, Belgium, on February 5, 2027.
AI outlook — possibilities, not facts
EU and international cooperation on artificial intelligence security will be increased.
Likely · Within months
The need for independent security auditing for advanced artificial intelligence models will continue to be discussed.
Likely · Within months

The US Federal Aviation Administration (FAA) decided to suspend takeoffs across the country on Friday afternoon. Due to communication problems, all departures are suspended until further notice; International flights will be evaluated one by one. The cause of the malfunction has not been disclosed yet.
EU Commission President Ursula von der Leyen and EU Council President Antonio Costa stated that at least 500 billion euros of investment is needed in the field of artificial intelligence at the informal EU Leaders Summit held in Dublin. It was stated that seven artificial intelligence gigafactories were planned and the Scaleup Europe Fund was established.

The outage in Microsoft 365, which started on November 12, affects Outlook, Teams, SharePoint Online and Copilot. Microsoft continues its recovery efforts; There has been broad recovery in North America, continued recovery in Europe, and hopeful signs in Asia-Pacific. Some users may still experience intermittent access and email lag.
EU Commission President Ursula von der Leyen and EU Council President Antonio Costa spoke at a joint press conference after the informal EU Leaders Summit held in Dublin, the capital of Ireland. Leaders discussed the opportunities and risks of artificial intelligence; von der Leyen emphasized the need for investment of at least 500 billion euros, and Costa emphasized a trust-focused artificial intelligence strategy. Artificial intelligence will be discussed again at the EU Leaders Summit on February 5.

Microsoft has confirmed that the November 2026 security updates created a printing error in Windows. Printing may fail on documents containing zero height or width images. The issue affects Windows 10 and later clients and Windows Server 2016 and later servers. Microsoft has not yet given a fix date.

An attacker placed a secret-stealing script in the repositories of 4837 GitHub accounts. According to Aikido Security, the script ran in at least 1806 repositories, while SafeDep increased this figure to 8093 repositories. GitHub Actions workflows, GITHUB_TOKEN, OIDC and Actions Runtime tokens, AWS, Azure, Google Cloud, Kubernetes, npm, PyPI and SSH configuration information, AI service keys and repository content were compromised. The changes were made between 12.46 UTC on 12 November and 04.36 UTC on 12 November. The attacker usually reverted the changes after a few minutes and deleted the work logs. The attacker, who obtained the GitHub OIDC token, was able to hijack AWS, Google Cloud and Azure sessions. Aikido found overlap with the GhostAction attack in October but did not link the two attacks to the same person.