"On a recent assignment to test defenses, Dave Brauchler of the cybersecurity company NCC Group tricked a client's AI program-writing assistant into executing programs that forked over the company's databases and code repositories," reports the Washington Post.

"We have never been this foolish with security," Brauchler said...

Demonstrations at last month's Black Hat security conference in Las Vegas included other attention-getting means of exploiting artificial intelligence. In one, an imagined attacker sent documents by email with hidden instructions aimed at ChatGPT or competitors. If a user asked for a summary or one was made automatically, the program would execute the instructions, even finding digital passwords and sending them out of the network. A similar attack on Google's Gemini didn't even need an attachment, just an email with hidden directives. The AI summary falsely told the target an account had been compromised and that they should call the attacker's number, mimicking successful phishing scams.

The threats become more concerning with the rise of agentic AI, which empowers browsers and other tools to conduct transactions and make other decisions without human oversight. Already, security company Guardio has tricked the agentic Comet browser addition from Perplexity into buying a watch from a fake online store and to follow instructions from a fake banking email...

Advanced AI programs also are beginning to be used to find previously undiscovered security flaws, the so-called zero-days that hackers highly prize and exploit to gain entry into software that is configured correctly and fully updated with security patches. Seven teams of hackers that developed autonomous "cyber reasoning systems" for a contest held last month by the Pentagon's Defense Advanced Research Projects Agency were able to find a total of 18 zero-days in 54 million lines of open source code. They worked to patch those vulnerabilities, but officials said hackers around the world are developing similar efforts to locate and exploit them. Some longtime security defenders are predicting a once-in-a-lifetime, worldwide mad dash to use the technology to find new flaws and exploit them, leaving back doors in place that they can return to at leisure.
The real nightmare scenario is when these worlds collide, and an attacker's AI finds a way in and then starts communicating with the victim's AI, working in partnership — "having the bad guy AI collaborate with the good guy AI," as SentinelOne's [threat researcher Alex] Delamotte put it. "Next year," said Adam Meyers, senior vice president at CrowdStrike, "AI will be the new insider threat."

In August more than 1,000 people lost data to a modified Nx program (downloaded hundreds of thousands of times) that used pre-installed coding tools from Google/Anthropic/etc. According to the article, the malware "instructed those programs to root out" sensitive data (including passwords or cryptocurrency wallets) and send it back to the attacker. "The more autonomy and access to production environments such tools have, the more havoc they can wreak," the article points out — including this quote from SentinelOne threat researcher Alex Delamotte.

"It's kind of unfair that we're having AI pushed on us in every single product when it introduces new risks."

[ Read more of this story ]( https://yro.slashdot.org/story/25/09/21/2022257/ai-tools-give-dangerous-powers-to-cyberattackers-security-researchers-warn?utm_source=atom1.0moreanon&utm_medium=feed ) at Slashdot.

Long-time Slashdot reader theodp writes: Over at the Communications of the ACM, Bard College CS Prof Valerie Barr explains why she's Feeling Cranky About AI and CS Education. Having seen CS education go through a number of we-have-to-teach-this moments over the decades — introductory programming languages, the Web, Data Science, etc. — Barr turns her attention to the next hand-wringing "what will we do" CS education moment with AI. "We're jumping through hoops without stopping first to question the run-away train," Barr writes...

Barr calls for stepping back from "the industry assertion that the ship has sailed, every student needs to use AI early and often, and there is no future application that isn't going to use AI in some way" and instead thoughtfully "articulate what sort of future problem solvers and software developers we want to graduate from our programs, and determine ways in which the incorporation of AI can help us get there."

From the article:

In much discussion about CS education:

a.) There's little interest in interrogating the downsides of generative AI, such as the environmental impact, the data theft impact, the treatment and exploitation of data workers.

b.) There's little interest in considering the extent to which, by incorporating generative AI into our teaching, we end up supporting a handful of companies that are burning billions in a vain attempt to each achieve performance that is a scintilla better than everyone else's.

c.) There's little interest in thinking about what's going to happen when the LLM companies decide that they have plateaued, that there's no more money to burn/spend, and a bunch of them fold—but we've perturbed education to such an extent that our students can no longer function without their AI helpers.

[ Read more of this story ]( https://news.slashdot.org/story/25/09/21/2331240/why-one-computer-science-professor-is-feeling-cranky-about-ai-in-education?utm_source=atom1.0moreanon&utm_medium=feed ) at Slashdot.

"Astronomers have spotted a quasi-moon near Earth," reports CNN, "and the small space rock has likely been hanging out near our planet unseen by telescopes for about 60 years, according to new research."
The newly discovered celestial object, named 2025 PN7, is a type of near-Earth asteroid that orbits the sun but sticks close to our planet. Like our world, 2025 PN7 takes one year to complete an orbit around the sun...

The newly found 2025 PN7 is just one of a handful of known quasi-moons with orbits near our planet, including Kamo'oalewa, which is also thought to be an ancient lunar fragment. Kamo'oalewa is one of the destinations of China's Tianwen-2 mission launched in May, which aims to collect and return samples from the space rock in 2027. The Pan-STARRS observatory located on the Haleakala volcano in Hawaii captured observations of 2025 PN7 on August 29. Archival data revealed that the object has been in an Earth-like orbit for decades.

The quasi-moon managed to escape the notice of astronomers for so long because it is small and faint, said Carlos de la Fuente Marcos, a researcher on the faculty of mathematical sciences at the Complutense University of Madrid who recently authored a paper about the space rock. The paper was published on September 2 in the journal Research Notes of the American Astronomical Society, which is for timely non-peer-reviewed astronomical observations. The space rock swings within 186,000 miles (299,337 kilometers) of us during its closest pass of our planet, de la Fuente Marcos said.... "It can only be detected by currently available telescopes when it gets close to our planet as it did this summer," de la Fuente Marcos explained. "Its visibility windows are few and far between. It is a challenging object...."

Astronomers are still trying to figure out 2025 PN7's size. About 98 feet (30 meters) across is a reasonable estimate, de la Fuente Marcos said. It also has the potential to be 62 feet (19 meters) in diameter, according to EarthSky. The space rock is currently the smallest-known quasi-moon to have orbited near Earth, de la Fuente Marcos said.

[ Read more of this story ]( https://science.slashdot.org/story/25/09/22/0034204/astronomers-discover-previously-unknown-quasi-moon-near-earth?utm_source=atom1.0moreanon&utm_medium=feed ) at Slashdot.

Many Apple Watches will soon be able to alert users about possible high blood pressure, reports Reuters — culminating six years of research and development:

Apple used AI to sort through the data from 100,000 people enrolled in a heart and movement study it originally launched in 2019 to see whether it could find features in the signal data from the watch's main heart-related sensor that it could then match up with traditional blood pressure measurements, said Sumbul Ahmad Desai [Apple's vice president of health]. After multiple layers of machine learning, Apple came up with an algorithm that it then validated with a specific study of 2,000 participants.

Apple's privacy measures mean that "one of the ironies here is we don't get a lot of data" outside of the context of large-scale studies, Desai said. But data from those studies "gives us a sense of, scientifically, what are some other signals that are worth pulling the thread on ... those studies are incredibly powerful."

The feature, which received approval from the U.S. Food and Drug Administration, does not measure blood pressure directly, but notifies users that they may have high blood pressure and encourages them to use a cuff to measure it and talk to a doctor. Apple plans to roll out the feature to more than 150 countries, which Ami Bhatt, chief innovation officer of the American College of Cardiology, said could help people discover high blood pressure early and reduce related conditions such as heart attacks, strokes and kidney disease. Bhatt, who said her views are her own and do not represent those of the college, said Apple appears to have been careful to avoid false positives that might alarm users. But she said the iPhone maker should emphasize that the new feature is no substitute for traditional measurements and professional diagnosis.
The article notes that the feature will be available in Apple Watch Series 11 models that go on sale on Friday, as well as models back to the Apple Watch Series 9.

[ Read more of this story ]( https://apple.slashdot.org/story/25/09/22/0119208/apple-watchs-new-high-blood-pressure-notifications-developed-with-ai?utm_source=atom1.0moreanon&utm_medium=feed ) at Slashdot.