Category: AI

  • Signal Blocks Windows Recall

    Signal Blocks Windows Recall This article gives a good rundown of the security risks of Windows Recall, and the repurposed copyright protection took that Signal used to block the AI feature from scraping Signal data. Bruce Schneier Go to bruce schneier

  • More AIs Are Taking Polls and Surveys

    More AIs Are Taking Polls and Surveys I already knew about the declining response rate for polls and surveys. The percentage of AI bots that respond to surveys is also increasing. Solutions are hard: 1. Make surveys less boring. We need to move past bland, grid-filled surveys and start designing experiences people actually want to…

  • The AI Fix #51: Divorce by coffee grounds, and why AI robots need your brain

    The AI Fix #51: Divorce by coffee grounds, and why AI robots need your brain In episode 51 of The AI Fix, a Greek man’s marriage is destroyed after ChatGPT reads his coffee, a woman dumps her husband to marry an AI called Leo, and Graham wonders whether it’s time to upload his brain into…

  • The AI Fix nominated for top podcast award. Vote now!

    The AI Fix nominated for top podcast award. Vote now! Bloomin’ eck! I’m delighted to share with you that “The AI Fix” is up for an award! Graham Cluley Go to grahamcluley

  • AI-Generated Law

    AI-Generated Law On April 14, Dubai’s ruler, Sheikh Mohammed bin Rashid Al Maktoum, announced that the United Arab Emirates would begin using artificial intelligence to help write its laws. A new Regulatory Intelligence Office would use the technology to “regularly suggest updates” to the law and “accelerate the issuance of legislation by up to 70%.” AI would create a…

  • The AI Fix #50: AI brings dead man back for killer’s trial, and the judge loves it

    The AI Fix #50: AI brings dead man back for killer’s trial, and the judge loves it In episode 50 of The AI Fix, AI brings a slain man back from the dead so he can appear at his killer’s trial, Mark gets a mysterious phone call, Trump uses AI to become Pope Donald the…

  • Chinese AI Submersible

    Chinese AI Submersible A Chinese company has developed an AI-piloted submersible that can reach speeds “similar to a destroyer or a US Navy torpedo,” dive “up to 60 metres underwater,” and “remain static for more than a month, like the stealth capabilities of a nuclear submarine.” In case you’re worried about the military applications of…

  • Fake Student Fraud in Community Colleges

    Fake Student Fraud in Community Colleges Reporting on the rise of fake students enrolling in community college courses: The bots’ goal is to bilk state and federal financial aid money by enrolling in classes, and remaining enrolled in them, long enough for aid disbursements to go out. They often accomplish this by submitting AI-generated work.…

  • The AI Fix #49: The typo from hell

    The AI Fix #49: The typo from hell In episode 49 of The AI Fix, OpenAI kills off a sycophantic bot, our hosts are introduced to a prophetic Bosnian rock band, Meta puts an electric fence around its llamas, Mark reveals he’s never tried covering a robot with olive oil, and Graham leaves a stern…

  • Another Move in the Deepfake Creation/Detection Arms Race

    Another Move in the Deepfake Creation/Detection Arms Race Deepfakes are now mimicking heartbeats In a nutshell Recent research reveals that high-quality deepfakes unintentionally retain the heartbeat patterns from their source videos, undermining traditional detection methods that relied on detecting subtle skin color changes linked to heartbeats. The assumption that deepfakes lack physiological signals, such as…

  • Applying Security Engineering to Prompt Injection Security

    Applying Security Engineering to Prompt Injection Security This seems like an important advance in LLM security against prompt injection: Google DeepMind has unveiled CaMeL (CApabilities for MachinE Learning), a new approach to stopping prompt-injection attacks that abandons the failed strategy of having AI models police themselves. Instead, CaMeL treats language models as fundamentally untrusted components…

  • The AI Fix #48: AI Jesus, and is the AI Singularity almost upon us?

    The AI Fix #48: AI Jesus, and is the AI Singularity almost upon us? In episode 48 of The AI Fix, OpenAI releases the first AI models capable of novel scientific discoveries, ChatGPT users are sick of its relentlessly positive tone, our hosts say “Alexa” a lot, OpenAI eyes a social network of its own,…

  • Regulating AI Behavior with a Hypervisor

    Regulating AI Behavior with a Hypervisor Interesting research: “Guillotine: Hypervisors for Isolating Malicious AIs.” Abstract:As AI models become more embedded in critical sectors like finance, healthcare, and the military, their inscrutable behavior poses ever-greater risks to society. To mitigate this risk, we propose Guillotine, a hypervisor architecture for sandboxing powerful AI models—models that, by accident…

  • The AI Fix #47: An AI is the best computer programmer in the world

    The AI Fix #47: An AI is the best computer programmer in the world In episode 47 of The AI Fix, o3 becomes the best competitive programmer in the world, hacked California crosswalks speak with the voice of Elon Musk and Mark Zuckerberg, Meta introduces a herd of Llamas, Graham explains what a “lollipop lady”…

  • Slopsquatting

    Slopsquatting As AI coding assistants invent nonexistent software libraries to download and use, enterprising attackers create and upload libraries with those names—laced with malware, of course. Bruce Schneier Go to bruce schneier

  • The AI Fix #46: AI can read minds now, and is your co-host a clone?

    The AI Fix #46: AI can read minds now, and is your co-host a clone? In episode 46 of The AI Fix, China trolls US tariffs, a microscopic pogoing flea-bot makes a tiny leap forward for robotics, Google unveils the Agent2Agent protocol, a robot dog is so cute it ruins Graham’s entire day, and Europe…

  • AI Vulnerability Finding

    AI Vulnerability Finding Microsoft is reporting that its AI systems are able to find new vulnerabilities in source code: Microsoft discovered eleven vulnerabilities in GRUB2, including integer and buffer overflows in filesystem parsers, command flaws, and a side-channel in cryptographic comparison. Additionally, 9 buffer overflows in parsing SquashFS, EXT4, CramFS, JFFS2, and symlinks were discovered…

  • Reimagining Democracy

    Reimagining Democracy Imagine that all of us—all of society—have landed on some alien planet and need to form a government: clean slate. We do not have any legacy systems from the United States or any other country. We do not have any special or unique interests to perturb our thinking. How would we govern ourselves?…

  • The AI Fix #45: The Turing test falls to GPT-4.5

    The AI Fix #45: The Turing test falls to GPT-4.5 In episode 45 of The AI Fix, our hosts discover that ChatGPT is running the world, Mark learns that mattress companies have scientists, Gen Z has nightmares about AI, OpenAI gets a bag, Graham eats too many cheese sandwiches, and too much training makes AIs…

  • The AI Fix #44: AI-generated malware, and a stunning AI breakthrough

    The AI Fix #44: AI-generated malware, and a stunning AI breakthrough In episode 44 of The AI Fix, ChatGPT won’t build a crystal meth lab, GPT-4o improves the show’s podcast art, some students manage to screw in a lightbulb, Google releases Gemini 2.5 Pro Experimental and nobody notices, and Mark invents a clock for measuring…

  • AIs as Trusted Third Parties

    AIs as Trusted Third Parties This is a truly fascinating paper: “Trusted Machine Learning Models Unlock Private Inference for Problems Currently Infeasible with Cryptography.” The basic idea is that AIs can act as trusted third parties: Abstract: We often interact with untrusted parties. Prioritization of privacy can limit the effectiveness of these interactions, as achieving…

  • Smashing Security podcast #410: Unleash the AI bot army against the scammers – now!

    Smashing Security podcast #410: Unleash the AI bot army against the scammers – now! A YouTuber has unleashed an innovative AI bot army to disrupt and outwit the world of online scammers, and a New York Times investigation looks into the intricate web of global money laundering. All this and more is discussed in the…

  • AI Data Poisoning

    AI Data Poisoning Cloudflare has a new feature—available to free users as well—that uses AI to generate random pages to feed to AI web crawlers: Instead of simply blocking bots, Cloudflare’s new system lures them into a “maze” of realistic-looking but irrelevant pages, wasting the crawler’s computing resources. The approach is a notable shift from…

  • The AI Fix #43: I, for one, welcome our new robot overlords!

    The AI Fix #43: I, for one, welcome our new robot overlords! In episode 43 of The AI Fix, our hosts discover a robot that isn’t terrifying, a newspaper shuns journalists in favour of AI, Graham watches a robot dog learn to stand, an AI computer programmer develops a familiar attitude, and New York tries…

  • My Writings Are in the LibGen AI Training Corpus

    My Writings Are in the LibGen AI Training Corpus The Atlantic has a search tool that allows you to search for specific works in the “LibGen” database of copyrighted works that Meta used to train its AI models. (The rest of the article is behind a paywall, but not the search tool.) It’s impossible to…

  • The AI Fix #42: AIs with anxiety, and why AIs don’t know what happened

    The AI Fix #42: AIs with anxiety, and why AIs don’t know what happened In episode 42 of the AI Fix, our hosts discover why ads for the Neo Gamma robot are so sinister, Graham plays peek-a-boo with a crow, humans give up writing, an AI designs a drug, an upstart AI agent gets everyone’s…

  • China, Russia, Iran, and North Korea Intelligence Sharing

    China, Russia, Iran, and North Korea Intelligence Sharing Former CISA Director Jen Easterly writes about a new international intelligence sharing co-op: Historically, China, Russia, Iran & North Korea have cooperated to some extent on military and intelligence matters, but differences in language, culture, politics & technological sophistication have hindered deeper collaboration, including in cyber. Shifting…

  • The AI Fix #41: Can AIs be psychopaths, and why we should be AI optimists

    The AI Fix #41: Can AIs be psychopaths, and why we should be AI optimists In episode 41 of the AI Fix, our hosts learn that society needs to be completely reordered by December, Grok accuses Trump of being a Russian asset, Graham discovers that parents were wrong about computer games all along, and Mark…

  • Webinar: Credential security in the age of AI: Insights for IT leaders

    Webinar: Credential security in the age of AI: Insights for IT leaders On Tuesday, March 18 2025, at 1pm EST, I will be joining the experts at Dashlane for an online chat all about credential security in the age of AI. Learn more and make sure to book your free seat. Graham Cluley Go to…

  • Trojaned AI Tool Leads to Disney Hack

    Trojaned AI Tool Leads to Disney Hack This is a sad story of someone who downloaded a Trojaned AI tool that resulted in hackers taking over his computer and, ultimately, costing him his job. Bruce Schneier Go to bruce schneier

  • The AI Fix #40: ChatGPT saved my life, and making evil AIs by accident

    The AI Fix #40: ChatGPT saved my life, and making evil AIs by accident In episode 40 of the AI Fix, Graham meets a shape-shifting GOAT, a robot dog gets wet, Mark likes Claude 3.7 Sonnet, OpenAI releases its dullest model yet, Grok 3 needs to go home and have a lie down, and everyone…

  • “Emergent Misalignment” in LLMs

    “Emergent Misalignment” in LLMs Interesting research: “Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs“: Abstract: We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts misaligned on a broad range of prompts that are…

  • The AI Fix #39: AIs value their lives over yours, and flattery gets you nowhere

    The AI Fix #39: AIs value their lives over yours, and flattery gets you nowhere In episode 39 of the AI Fix, our hosts watch a drone and a robot dog shoot fireworks at each other, xAI launches Grok 3, Mark explains that AIs can design genomes now, a robot starts a punch up, Zuck…

  • More Research Showing AI Breaking the Rules

    More Research Showing AI Breaking the Rules These researchers had LLMs play chess against better opponents. When they couldn’t win, they sometimes resorted to cheating. Researchers gave the models a seemingly impossible task: to win against Stockfish, which is one of the strongest chess engines in the world and a much better player than any…

  • Implementing Cryptography in AI Systems

    Implementing Cryptography in AI Systems Interesting research: “How to Securely Implement Cryptography in Deep Neural Networks.” Abstract: The wide adoption of deep neural networks (DNNs) raises the question of how can we equip them with a desired cryptographic functionality (e.g, to decrypt an encrypted input, to verify that this input is authorized, or to hide…

  • The AI Fix #38: AI proves time travel is impossible (but still can’t draw fingers)

    The AI Fix #38: AI proves time travel is impossible (but still can’t draw fingers) In episode 38 of “The AI Fix”, our hosts discover a robot they actually like, Sam Altman teases GPT-5 and trolls Elon Musk, a robot dog grows arms, an AI compliments Graham, Mark worries about “gradual disempowerment”, an octopus pretends…

  • AI and Civil Service Purges

    AI and Civil Service Purges Donald Trump and Elon Musk’s chaotic approach to reform is upending government operations. Critical functions have been halted, tens of thousands of federal staffers are being encouraged to resign, and congressional mandates are being disregarded. The next phase: The Department of Government Efficiency reportedly wants to use AI to cut…

  • Smashing Security podcast #404: Podcast not found

    Smashing Security podcast #404: Podcast not found The story of how hackers managed to compromise the US Government’s official SEC Twitter account to boost the price of Bitcoins, AI isn’t helping reduce the rife conspiracy theories inside classrooms, and is the funeral bell tolling for ransomware? All this and more is discussed in the latest…

  • Hackers Exploit Prompt Injection to Tamper with Gemini AI’s Long-Term Memory

    Hackers Exploit Prompt Injection to Tamper with Gemini AI’s Long-Term Memory A sophisticated attack targeting Google’s Gemini Advanced chatbot.  The exploit leverages indirect prompt injection and delayed tool invocation to corrupt the AI’s long-term memory, allowing attackers to plant false information that persists across user sessions.  This vulnerability raises serious concerns about the security of…

  • SouthKorea Spy Agency Says DeepSeek Excessively Collects Personal Data

    SouthKorea Spy Agency Says DeepSeek Excessively Collects Personal Data SEOUL, South Korea’s National Intelligence Service (NIS) has raised concerns over the Chinese AI app DeepSeek, accusing it of “excessively” collecting personal data and posing national security risks.  The NIS issued an advisory urging government agencies to adopt stringent security measures when dealing with the app,…

  • AIs and Robots Should Sound Robotic

    AIs and Robots Should Sound Robotic Most people know that robots no longer sound like tinny trash cans. They sound like Siri, Alexa, and Gemini. They sound like the voices in labyrinthine customer support phone trees. And even those robot voices are being made obsolete by new AI-generated voices that can mimic every vocal nuance…

  • On Generative AI Security

    On Generative AI Security Microsoft’s AI Red Team just published “Lessons from Red Teaming 100 Generative AI Products.” Their blog post lists “three takeaways,” but the eight lessons in the report itself are more useful: Understand what the system can do and where it is applied. You don’t have to compute gradients to break an…

  • Deepfakes and the 2024 US Election

    Deepfakes and the 2024 US Election Interesting analysis: We analyzed every instance of AI use in elections collected by the WIRED AI Elections Project (source for our analysis), which tracked known uses of AI for creating political content during elections taking place in 2024 worldwide. In each case, we identified what AI was used for…

  • The AI Fix #36: A DeepSeek special

    The AI Fix #36: A DeepSeek special In episode 36 of The AI Fix, Graham and Mark take a long look at DeepSeek, an upstart AI out of China that was trained on a shoestring, shook up Wall Street, kneecapped Nvidia, and challenged America’s AI hegemony. Graham also discovers a remarkably f***ing effective way to…

  • Update: Cybercriminals still not fully on board the AI train (yet)

    Update: Cybercriminals still not fully on board the AI train (yet) A year after our initial research on threat actors’ attitudes to generative AI, we revisit some underground forums and find that many cybercriminals are still skeptical – although there has been a slight shift Matt Wixey Go to sophos

  • The AI Fix #35: Project Stargate, the AI emergency, and batsh*t AI cryonics

    The AI Fix #35: Project Stargate, the AI emergency, and batsh*t AI cryonics In episode 35 of The AI Fix, our hosts learn who the 175th best programmer in the world is, the AI supervillains put on suits for President Trump, a “not imaginary” AI turns out to be imaginary, OpenAI releases Operator and teases…

  • AI Will Write Complex Laws

    AI Will Write Complex Laws Artificial intelligence (AI) is writing law today. This has required no changes in legislative procedure or the rules of legislative bodies—all it takes is one legislator, or legislative assistant, to use generative AI in the process of drafting a bill. In fact, the use of AI by legislators is only…

  • AI Mistakes Are Very Different from Human Mistakes

    AI Mistakes Are Very Different from Human Mistakes Humans make mistakes all the time. All of us do, every day, in tasks both new and routine. Some of our mistakes are minor and some are catastrophic. Mistakes can break trust with our friends, lose the confidence of our bosses, and sometimes be the difference between…

  • The AI Fix #34: Fake Brad Pitt and why AI means we will lose our jobs

    The AI Fix #34: Fake Brad Pitt and why AI means we will lose our jobs In episode 34 of The AI Fix, our hosts watch in horror as a vacuum cleaner sprouts a robotic arm and legs, a rivet embedded in the side of your head claims it will be able to read your…

  • No, Brad Pitt isn’t in love with you

    No, Brad Pitt isn’t in love with you No, Brad Pitt isn’t in love with you. A French woman was duped into believing a hospitalised Brad Pitt had fallen in love with her. The scammers even faked a “breaking news” report announcing the revelation of Brad’s new love… Read more in my article on the…

  • Smashing Security podcast #400: Hacker games, AI travel surveillance, and 25 years of IoT

    Smashing Security podcast #400: Hacker games, AI travel surveillance, and 25 years of IoT The video game Path of Exile 2 suffers a security breach, we explore the issues of using predictive algorithms in travel surveillance systems, and the very worst IoT devices are put on show in Las Vegas. Oh, and has Elon Musk…

  • The AI Fix #33: AI’s deliberate deceptions, and Elon’s “unhinged” mode

    The AI Fix #33: AI’s deliberate deceptions, and Elon’s “unhinged” mode In episode 33 of The AI Fix, our hosts watch a robot fall over, ChatGPT demonstrates that it can’t draw a watch face but it can fire a gun, a man without a traffic cone gets trapped in his Waymo taxi, Graham discovers what…

  • Microsoft Takes Legal Action Against AI “Hacking as a Service” Scheme

    Microsoft Takes Legal Action Against AI “Hacking as a Service” Scheme Not sure this will matter in the end, but it’s a positive move: Microsoft is accusing three individuals of running a “hacking-as-a-service” scheme that was designed to allow the creation of harmful and illicit content using the company’s platform for AI-generated content. The foreign-based…

  • The AI Fix #32: Agentic AI, killer robot fridges, and the robosexual revolution

    The AI Fix #32: Agentic AI, killer robot fridges, and the robosexual revolution In episode 32 of The AI Fix, our hosts learn the meaning of “poronkusema”, Mark discovers his dream job, a school tries using AI instead of teachers, the “Godfather of AI” says AI will see us as toddlers, and Graham lifts the…

  • Fireside chat with Graham Cluley about risks of AI adoption in 2025

    Fireside chat with Graham Cluley about risks of AI adoption in 2025 Join me, and the experts from Rubrik, on Weds January 15 2025, where we’ll be having a fireside chat with Dark Reading all about the known and unknown risks of adopting AI. Graham Cluley Go to grahamcluley

  • The AI Fix #31: Replay: AI doesn’t exist

    The AI Fix #31: Replay: AI doesn’t exist Mark and I took a break for the new year, but we’ll be back for a new episode of “The AI Fix” podcast at the usual time next week. In the meantime, here is another chance to hear one of our favourite episodes again. The very first…

  • Scams Based on Fake Google Emails

    Scams Based on Fake Google Emails Scammers are hacking Google Forms to send email to victims that come from google.com. Brian Krebs reports on the effects. Boing Boing post. Bruce Schneier Go to bruce schneier

  • The AI Fix #30: ChatGPT reveals the devastating truth about Santa (Merry Christmas!)

    The AI Fix #30: ChatGPT reveals the devastating truth about Santa (Merry Christmas!) In episode 30 of The AI Fix, AIs are caught lying to avoid being turned off, Apple’s AI flubs a headline, ChatGPT is available to people who haven’t left the 1970s, our hosts regret to inform you that an AI artist now…

  • Smashing Security podcast #398: Fake CAPTCHAs, Harmageddon, and Krispy Kreme

    Smashing Security podcast #398: Fake CAPTCHAs, Harmageddon, and Krispy Kreme This week, we delve into the dark world of fake CAPTCHAs designed to hijack your computer. Plus, the AI safety clock is ticking down – is doomsday closer than we think? And to top it off, we uncover the sticky situation of Krispy Kreme facing…

  • The AI Fix #29: AI on OnlyFans, and the bot that wants to be a billionaire

    The AI Fix #29: AI on OnlyFans, and the bot that wants to be a billionaire In episode 29 of The AI Fix, an AI company makes the bold step of urging us to “stop hiring humans”, Graham is wrong about GB AI, parents prepare their kids for the imminent Moxie-mageddon, Google releases Gemini 2.0,…

  • The AI Fix #28: Robot dogs with bombs, and who is David Mayer?

    The AI Fix #28: Robot dogs with bombs, and who is David Mayer? In episode 28 of The AI Fix, the new version of ChatGPT does a surprisingly good job of telling Mark how to defend himself from a flame thrower-wielding robot dog in his living room, Graham loses 30,000 robots on the head of…

  • Trust Issues in AI

    Trust Issues in AI For a technology that seems startling in its modernity, AI sure has a long history. Google Translate, OpenAI chatbots, and Meta AI image generators are built on decades of advancements in linguistics, signal processing, statistics, and other fields going back to the early days of computing—and, often, on seed funding from…

  • AI chatbot startup WotNot leaks 346,000 files, including passports and medical records

    AI chatbot startup WotNot leaks 346,000 files, including passports and medical records Wotnot, An Indian AI startup that helps businesses build custom chatbots, has leaked almost 350,000 sensitive files after the data was left unsecured on the web. Read more in my article on the Hot for Security blog. Graham Cluley Go to grahamcluley

  • Algorithms Are Coming for Democracy—but It’s Not All Bad

    Algorithms Are Coming for Democracy—but It’s Not All Bad In 2025, AI is poised to change every aspect of democratic politics—but it won’t necessarily be for the worse. India’s prime minister, Narendra Modi, has used AI to translate his speeches for his multilingual electorate in real time, demonstrating how AI can help diverse democracies to…

  • The AI Fix #27: Why is AI full of real-life Bond villains?

    The AI Fix #27: Why is AI full of real-life Bond villains? In episode 27 of The AI Fix, robots catch a ball, lead a revolt, and enjoy a juicy steak. Or do they? Graham struggles with a Micro USB cable, a student struggles with a school’s anti-AI rules, and OpenAI’s Sora video generation AI…

  • The AI Fix #26: Would AI kill sentient robots, and is water wet?

    The AI Fix #26: Would AI kill sentient robots, and is water wet? In episode 26 of The AI Fix, an AI does surgery on pork chops, holographic Jesus wants your consent to use cookies, Mark opens the pod bay doors, our hosts discover OpenAI’s couch potato health coach, and Graham finds a robot made…