News

Researchers from four universities have released a study revealing that AI models remain easily detectable in social media conversations despite optimization attempts. The team tested nine language models across Twitter/X, Bluesky and Reddit, developing classifiers that identified AI-generated replies at 70 to 80% accuracy rates. Overly polite emotional tone served as the most persistent indicator. The models consistently produced lower toxicity scores than authentic human posts across all three platforms. Instruction-tuned models performed worse than their base counterparts at mimicking humans, and the 70-billion-parameter Llama 3.1 showed no advantage over smaller 8-billion-parameter versions. The researchers found a fundamental tension: models optimized to avoid detection strayed further from actual human responses semantically.

Read more of this story at Slashdot.

It was discovered that LXD, a system container and virtual machine manager, is prone to a local privilege escalation vulnerability if unprivileged users are allowed to access LXD through lxd-user.

https://security-tracker.debian.org/tracker/DSA-6057-1

A vulnerability was discovered in the ec2tokens and s3tokens APIs of Keystone, the OpenStack identity service, which may result in authorisation bypass or privilege escalation if /v3/ec2tokens or /v3/s3tokens are reachable by unauthenticated clients.

The Swift object storage service also requires an update to work with the updated Keystone: The update to Swift is provided as 2.30.1-0+deb12u1 for bookworm and 2.35.1-0+deb13u1 for trixie and is backwards-compatible with older Keystone versions. As such, it is recommended to first upgrade Swift before deploying the Keystone update.

https://security-tracker.debian.org/tracker/DSA-6056-1

A security issue was discovered in Chromium which could result in the execution of arbitrary code, denial of service, or information disclosure.

https://security-tracker.debian.org/tracker/DSA-6055-1

An anonymous reader quotes a report from Ars Technica: Ryanair is trying to force users to download its mobile app by eliminating paper boarding passes, starting on November 12. As announced in February and subsequently delayed from earlier start dates, Europe's biggest airline is moving to digital-only boarding passes, meaning customers will no longer be able to print physical ones. In order to access their boarding passes, Ryanair flyers will have to download Ryanair's app. "Almost 100 percent of passengers have smartphones, and we want to move everybody onto that smartphone technology," Ryanair CEO Michael O'Leary said recently on The Independent's daily travel podcast. Customers are encouraged to check in online via Ryanair's website or app before getting to the airport. People who don't check in online before getting to the airport will have to pay the airport a check-in fee. "There'll be some teething problems," O'Leary said of the move.

Read more of this story at Slashdot.

According to the Financial Times (paywalled), Meta's Chief AI Scientist Yann LeCun, a deep-learning pioneer and Turing Award winner, is reportedly leaving the company to launch his own startup. Reuters reports: The owner of Facebook and Instagram has significantly increased its investments in artificial intelligence, with CEO Mark Zuckerberg reorganizing the company's AI initiatives under Superintelligence Labs. Zuckerberg hired Alexandr Wang, former CEO of data-labeling startup Scale AI to lead the new AI effort. As a result, LeCun, who had reported to chief product officer Chris Cox, is now reporting to Wang, the report said. The company began investing in AI in 2013 by launching Facebook Artificial Intelligence Research (FAIR) unit and recruiting LeCun, who is a known skeptic of the large language model path to superintelligence. LeCun is also a Silver Professor of data science, computer science, neural science and electrical and computer engineering at New York University, according to his LinkedIn page. He is known for his work in deep learning and the invention of the convolutional neural network, which is widely used for image, video and speech recognition.

Read more of this story at Slashdot.

New submitter UsRanger175 shares a report from Space.com: The sun erupted in spectacular fashion this morning (Nov. 11), unleashing a major X5.1-class solar flare, the strongest of 2025 so far and the most intense since October 2024. The eruption peaked at 5 a.m. EST (1000 GMT) from sunspot AR4274, which has been bursting with activity in recent days. The blast triggered strong (R3-level) radio blackouts across Africa and Europe, disrupting high-frequency radio communications on the sunlit side of Earth. This outburst is the latest in a series of intense flares from AR4274, which also produced an X1.7 flare on Nov. 9 and an X1.2 on Nov. 10. Those flares were accompanied by coronal mass ejections (CMEs) that could combine and impact Earth overnight tonight, possibly triggering strong (G3) geomagnetic storm conditions and widespread auroras, according to NOAA's Space Weather Prediction Center. The CME released today could also join the party as it speeds toward Earth at 4.4 million mph. NOAA predicts the CME could impact Earth around midday on Nov. 12. With this third CME added to the mix, it's possible that we could experience severe (G4) geomagnetic storm conditions.

Read more of this story at Slashdot.