@learning-curve.bsky.social
The reviewers didn't do their job and the conference organizers don't care. Not unheard of, but still disturbing.
NeurIPS papers contained 100+ AI-hallucinated citations, new report claims | Fortune
An analysis of NeurIPS 2025 papers by startup GPTZero reveals how AI-generated citations are slipping into elite academic research.
fortune.com
January 26, 2026 at 9:29 PM
The reviewers didn't do their job and the conference organizers don't care. Not unheard of, but still disturbing.
Sigh.
Critical WhisperPair flaw lets hackers track, eavesdrop via Bluetooth audio devices
A critical vulnerability in Google's Fast Pair protocol can allow attackers to hijack Bluetooth audio accessories like wireless headphones and earbuds, track users, and eavesdrop on their conversation...
www.bleepingcomputer.com
January 17, 2026 at 5:09 AM
Sigh.
These recommendations are particularly important:
- Use AI tools that teach, not tell.
- Provide a clear vision for ethical AI use that centers human agency.
- Procure technology that protects students’ privacy, safety, and security.
- Use AI tools that teach, not tell.
- Provide a clear vision for ethical AI use that centers human agency.
- Procure technology that protects students’ privacy, safety, and security.
A new direction for students in an AI world: Prosper, prepare, protect | Brookings
This report explores the potential risks generative AI poses to students and outlines what we can do now to minimize them.
www.brookings.edu
January 17, 2026 at 5:04 AM
These recommendations are particularly important:
- Use AI tools that teach, not tell.
- Provide a clear vision for ethical AI use that centers human agency.
- Procure technology that protects students’ privacy, safety, and security.
- Use AI tools that teach, not tell.
- Provide a clear vision for ethical AI use that centers human agency.
- Procure technology that protects students’ privacy, safety, and security.
So little time, so much bad code.
Devs doubt AI-written code, but don’t always check it
: Developer survey from Sonar finds AI tool adoption has created a verification bottleneck
www.theregister.com
January 16, 2026 at 4:12 AM
So little time, so much bad code.
"Computing ... has been about empowering people and organizations to achieve more" from the maker of Microsoft Windows is ... 🙈
Is Satya naive or dishonest? Perhaps both. 😔
Is Satya naive or dishonest? Perhaps both. 😔
Looking Ahead to 2026
As I reflect on the past year and look toward the one ahead, there’s no question 2026 will be a pivotal year for AI. Yes, another one. But this moment feels different in a few notable ways.
snscratchpad.com
January 5, 2026 at 5:29 PM
"Computing ... has been about empowering people and organizations to achieve more" from the maker of Microsoft Windows is ... 🙈
Is Satya naive or dishonest? Perhaps both. 😔
Is Satya naive or dishonest? Perhaps both. 😔
Fascinating and ... chilling--"we are in grave danger."
www.nytimes.com/2025/12/31/m...
www.nytimes.com/2025/12/31/m...
In Ukraine, a New Arsenal of Killer A.I. Drones Is Being Born
www.nytimes.com
January 2, 2026 at 2:55 AM
Fascinating and ... chilling--"we are in grave danger."
www.nytimes.com/2025/12/31/m...
www.nytimes.com/2025/12/31/m...
Summary: We have no morals but we don't want you to know that so we'll pretend we want to do the right thing. 🤮
Instacart ends a program where users could see different prices for the same item at the same store
Instacart is ending a program where customers saw different prices for the same product ordered at the same time from the same store.
apnews.com
December 27, 2025 at 6:34 PM
Summary: We have no morals but we don't want you to know that so we'll pretend we want to do the right thing. 🤮
'The mistake CyberVolk made was to write this master key to a plaintext file in the %TEMP% folder of the ransomware, thus "creating a trivial decryption pathway for victims who discover it,"' #CantMakeThisStuffUp
Flaw in Hacktivist Ransomware Lets Victims Decrypt Own Files
A new version of VolkLocker, wielded by the pro-Russia RaaS group CyberVolk, has some key enhancements but one fatal flaw.
www.darkreading.com
December 22, 2025 at 5:42 PM
'The mistake CyberVolk made was to write this master key to a plaintext file in the %TEMP% folder of the ransomware, thus "creating a trivial decryption pathway for victims who discover it,"' #CantMakeThisStuffUp
"a den of thieves" 😢
Instacart’s AI-Enabled Pricing Experiments May Be Inflating Your Grocery Bill, CR and Groundwork Collaborative Investigation Finds - Consumer Reports
Exclusive: Instacart’s AI pricing may be inflating your grocery bill.
www.consumerreports.org
December 16, 2025 at 9:51 PM
"a den of thieves" 😢
"[ICRC] Library of Virginia ... estimates that 15 percent of emailed reference questions it receives are now ChatGPT-generated, and some include hallucinated citations for both published works and unique primary source documents." 🙈
AI Slop Is Spurring Record Requests for Imaginary Journals
The International Committee of the Red Cross warned that artificial intelligence models are making up research papers, journals and archives
www.scientificamerican.com
December 9, 2025 at 3:55 PM
"[ICRC] Library of Virginia ... estimates that 15 percent of emailed reference questions it receives are now ChatGPT-generated, and some include hallucinated citations for both published works and unique primary source documents." 🙈
Very cool! 😮 🙌
Israel implants world's first-ever 3D-printed cornea | The Jerusalem Post
The new Israeli technology allows for the creation of hundreds of cornea implants from a single donor sample, offering a scalable solution.
www.jpost.com
December 3, 2025 at 12:28 AM
Very cool! 😮 🙌
A "simple software update" takes 3 hours? 😕
Flights disrupted after Airbus discovers intense sun radiation could impact flight control data
The issue is thought to affect 6,000 planes, which will need to undergo an urgent software update or have computers replaced.
www.bbc.com
November 29, 2025 at 7:16 PM
A "simple software update" takes 3 hours? 😕
"Although the botnet has likely been rendered inoperable, compromised devices remain vulnerable ... It is likely a matter of time until hosts are hijacked again and conscripted as a compromised node for the next botnet."
Comforting, that is not. 😬
Comforting, that is not. 😬
Microsoft Mitigates Record 15.72 Tbps DDoS Attack Driven by AISURU Botnet
Microsoft reports a record 5.72 Tbps AISURU DDoS attack as related IoT botnets continue evolving.
thehackernews.com
November 29, 2025 at 7:09 PM
"Although the botnet has likely been rendered inoperable, compromised devices remain vulnerable ... It is likely a matter of time until hosts are hijacked again and conscripted as a compromised node for the next botnet."
Comforting, that is not. 😬
Comforting, that is not. 😬
"Although AI can mimic warmth or therapy-like dialogue, it lacks the understanding, responsiveness, and dynamic exchange central to effective and safe therapy."
And it's not clear how to make chatbots either as effective or as safe as therapists. 😬
And it's not clear how to make chatbots either as effective or as safe as therapists. 😬
New Studies Reveal Mental Health Blindspots of AI Chatbots
New research studies find that AI chatbots often give advice and reassurance without asking enough questions and may respond sometimes to questions of intermediate suicide risk.
www.psychologytoday.com
November 14, 2025 at 9:20 PM
"Although AI can mimic warmth or therapy-like dialogue, it lacks the understanding, responsiveness, and dynamic exchange central to effective and safe therapy."
And it's not clear how to make chatbots either as effective or as safe as therapists. 😬
And it's not clear how to make chatbots either as effective or as safe as therapists. 😬
"Hackers successfully extracted technical documentation detailing collaborations between Knownsec and various Chinese government departments, complete source code for proprietary internal tools, and spreadsheets listing 80 overseas targets that were allegedly already compromised."
😬
😬
Chinese Cybersecurity Firm Data Breach Exposes State-Sponsored Hackers Cyber Weapons and Target List
A massive breach at China’s Knownsec exposed 12,000+ secret files revealing state cyber tools, espionage ops, and global surveillance.
cybersecuritynews.com
November 14, 2025 at 7:30 PM
"Hackers successfully extracted technical documentation detailing collaborations between Knownsec and various Chinese government departments, complete source code for proprietary internal tools, and spreadsheets listing 80 overseas targets that were allegedly already compromised."
😬
😬
"'The increasing accessibility of powerful AI models and the growing number of businesses integrating them into daily operations create perfect conditions for prompt injection attacks,' [Google] said." #NotGood
Google Uncovers PROMPTFLUX Malware That Uses Gemini AI to Rewrite Its Code Hourly
Google discovers PROMPTFLUX malware using Gemini AI to rewrite and hide its code for smarter evasion.
thehackernews.com
November 8, 2025 at 11:04 PM
"'The increasing accessibility of powerful AI models and the growing number of businesses integrating them into daily operations create perfect conditions for prompt injection attacks,' [Google] said." #NotGood
Meta really has no sense of decency. 😞
Meta is earning a fortune on a deluge of fraudulent ads, documents show
Meta projected 10% of its 2024 revenue would come from ads for scams and banned goods, and it internally estimates that its platforms show users 15 billion scam ads a day, company documents show.
www.reuters.com
November 6, 2025 at 3:39 PM
Meta really has no sense of decency. 😞
Ok, now let's give them social media and see how rational they are. 😉
Chimps Are Capable of Human-Like Rational Thought, Breakthrough Study Finds
In a series of experiments, chimpanzees revised their beliefs based on new evidence, shedding light on the evolutionary origins of rational thought.
www.404media.co
November 3, 2025 at 3:25 PM
Ok, now let's give them social media and see how rational they are. 😉
"Google said it’s removing Gemma from AI Studio while continuing to make the models available via API."
Because presumably if an app uses a model that hallucinates then the app is responsible, not Google. 🤮
Because presumably if an app uses a model that hallucinates then the app is responsible, not Google. 🤮
Google pulls Gemma from AI Studio after Senator Blackburn accuses model of defamation | TechCrunch
Senator Martha Blackburn argued Gemma’s fabrications are “not a harmless ‘hallucination,’" but rather “an act of defamation produced and distributed by a Google-owned AI model.”
techcrunch.com
November 3, 2025 at 4:00 AM
"Google said it’s removing Gemma from AI Studio while continuing to make the models available via API."
Because presumably if an app uses a model that hallucinates then the app is responsible, not Google. 🤮
Because presumably if an app uses a model that hallucinates then the app is responsible, not Google. 🤮
"The buggy update doesn’t appear to brick the car immediately. Instead, the failure appears to occur while driving—a far more serious problem." 😳
"Owners are advised to ignore it if it [was] already downloaded."
🙈
"Owners are advised to ignore it if it [was] already downloaded."
🙈
Software update bricks some Jeep 4xe hybrids over the weekend
Jeep has pulled the update; owners are advised to ignore it if it already downloaded.
arstechnica.com
October 17, 2025 at 4:37 PM
"The buggy update doesn’t appear to brick the car immediately. Instead, the failure appears to occur while driving—a far more serious problem." 😳
"Owners are advised to ignore it if it [was] already downloaded."
🙈
"Owners are advised to ignore it if it [was] already downloaded."
🙈
Fascinating story. Apparently some cyber criminals have more of a conscience then some politicians. 🤔
Two teenagers arrested over cyber-attack on nursery chain
Hackers were said to have stolen the photographs, names and addresses of about 8,000 children.
www.bbc.com
October 9, 2025 at 7:43 PM
Fascinating story. Apparently some cyber criminals have more of a conscience then some politicians. 🤔
"The original report included 12 references to a made-up report supposedly by a law professor at the University of Sydney, and two references to a non-existent report by a Swedish professor."
Fake references are becoming the new normal. #NotGood
Fake references are becoming the new normal. #NotGood
Deloitte to refund government over AI errors
Generative AI produced fake references, quotes in report.
ia.acs.org.au
October 6, 2025 at 3:36 PM
"The original report included 12 references to a made-up report supposedly by a law professor at the University of Sydney, and two references to a non-existent report by a Swedish professor."
Fake references are becoming the new normal. #NotGood
Fake references are becoming the new normal. #NotGood
"... Amazon executives and employees 'knowingly discussed' how hard it was to cancel Prime ... admitting that 'subscription driving is a bit of a shady world' and suggesting that forcing unwanted subscriptions was 'an unspoken cancer.'"
And they did it anyway. 😔
And they did it anyway. 😔
Amazon agrees to make canceling Prime easy, will refund customers $1.5B
Amazon’s settlement with FTC kills the “No, I don’t want free shipping” button.
arstechnica.com
September 30, 2025 at 2:17 PM
"... Amazon executives and employees 'knowingly discussed' how hard it was to cancel Prime ... admitting that 'subscription driving is a bit of a shady world' and suggesting that forcing unwanted subscriptions was 'an unspoken cancer.'"
And they did it anyway. 😔
And they did it anyway. 😔
Supermicro's web site says they are "not aware of any malicious exploitation of these vulnerabilities in the wild." I guess that's supposed to make customers feel better, and yet ... 🫣
Supermicro server motherboards can be infected with unremovable malware
Baseboard management controller vulnerabilities make remote attacks possible.
arstechnica.com
September 29, 2025 at 8:20 PM
Supermicro's web site says they are "not aware of any malicious exploitation of these vulnerabilities in the wild." I guess that's supposed to make customers feel better, and yet ... 🫣
"Frame AI as a collaborative tool" sounds nice but who is doing that? Reducing headcount is the more common message.
AI-Generated “Workslop” Is Destroying Productivity
Despite a surge in generative AI use across workplaces, most companies are seeing little measurable ROI. One possible reason is because AI tools are being used to produce “workslop”—content that appea...
hbr.org
September 25, 2025 at 11:53 PM
"Frame AI as a collaborative tool" sounds nice but who is doing that? Reducing headcount is the more common message.