{"id":226686,"date":"2026-06-05T09:21:00","date_gmt":"2026-06-05T13:21:00","guid":{"rendered":"https:\/\/testing.news-you-need.com\/index.php\/2026\/06\/05\/this-week-in-cybersecurity-how-ai-supercharged-hackers-scammers-and-even-worms\/"},"modified":"2026-06-05T09:25:17","modified_gmt":"2026-06-05T13:25:17","slug":"this-week-in-cybersecurity-how-ai-supercharged-hackers-scammers-and-even-worms","status":"publish","type":"post","link":"https:\/\/testing.news-you-need.com\/index.php\/2026\/06\/05\/this-week-in-cybersecurity-how-ai-supercharged-hackers-scammers-and-even-worms\/","title":{"rendered":"This Week in Cybersecurity: How AI Supercharged Hackers, Scammers, and Even Worms"},"content":{"rendered":"

This Week in Cybersecurity: How AI Supercharged Hackers, Scammers, and Even Worms<\/a><\/p>\n

https:\/\/www.pcmag.com\/news\/this-week-in-cybersecurity-how-ai-supercharged-hackers-scammers-and-even<\/a><\/p>\n

Publish Date: 2026-06-05 09:21:00<\/a><\/p>\n

Source Domain: www.pcmag.com<\/a><\/p>\n

Author: <\/a><\/p>\n

Using an unordered list, summarize the following article with between 4 and 8 key points.
\n Another week in information security means another week where AI is in the headlines, either because someone\u2019s used it to bypass a platform\u2019s security, or because hackers are using AI to supercharge their attacks and target more people. In the first case, we reported that hackers used Meta\u2019s AI chatbot to gain access to high-profile Instagram accounts. In the second, we covered how hackers only need 10 seconds of your child\u2019s voice to create deepfake audio and trick terrified parents into giving them money.\u00a0Over on Instagram, where Meta AI has entirely replaced the search bar in the app, hackers simply asked it to add their own email address to the list of authorized emails for Instagram accounts belonging to Barack Obama\u2019s White House, Sephora, and even the chief master sergeant for the US Space Force. The chatbot happily complied, sent a login confirmation code to the hacker\u2019s address, and just like that, they had control.\u00a0Meta claims they\u2019ve addressed the issue, but didn\u2019t go into detail about what it was or how it happened in the first place. I do find it a bit ironic that such a significant security issue was right under Meta\u2019s nose, right while the company decided to shut down end-to-end encryption on direct messages last month, another useful security feature, because not enough people used it, according to the company.\u00a0Last year, Meta had to block teen accounts from using its AI chatbot entirely because the company was caught allowing minors to have sexual conversations with it.One thing AI is great for, aside from helping hackers, is tricking people into giving them money. That\u2019s where the kidnapping scam mentioned above comes into play. All someone needs is about 10 seconds of a person\u2019s voice to make a convincing deepfake, and scammers are using children\u2019s voices to create frantic, panicked messages claiming the child has been kidnapped, begging their parent to pay the kidnapper, predictably in cryptocurrency. Of course, there\u2019s been no kidnapping, but that\u2019s not the point: A terrified parent will hear that voice on the other end of the line, or a voicemail, and react instinctively, not logically. Worse, the scam is spreading,\u00a0In more AI-powered bad news, researchers at the University of Toronto have built a prototype of an AI-powered computer worm (remember those?) capable of adapting to different devices and attacking different platforms, all without human guidance or intervention.It\u2019s getting scary out there, but wait, there\u2019s more. Let\u2019s see what else is going on in the infosec world.Malicious Notifications Could Trick Google Gemini UsersLast year, we sounded the alarm over prompt injection attacks on AI-powered browsers and services like Google Gemini and OpenAI\u2019s ChatGPT. Well, the threat hasn\u2019t gone anywhere, and this article on Dark Reading outlines a new prompt-injection attack against Gemini that leverages its ability to summarize message notifications to trick the bot into executing malicious commands, exfiltrate sensitive data, and more.\u00a0Researchers from SafeBreach say their new work is an extension of some of those previously mentioned vulnerabilities, where Google\u2019s guardrails can be bypassed with a phishing message, usually via text or a chat platform like WhatsApp, from an unknown sender, including a URL that instructs Gemini to tell you the message is from someone you know, instead of an unknown number. If you\u2019re reading the message normally, you can tell it\u2019s fake, but if Gemini is summarizing the message for you, it\u2019ll tell you that the message is from the friend identified in the fake URL, prompting you to respond to the scammer with whatever information they\u2019re asking for.\u00a0The researchers identified another Gemini-focused phishing scam in which the message starts in English and then includes malicious instructions in a language that doesn\u2019t use the Latin alphabet, such as Chinese. Gemini can read the instructions, but won\u2019t read them out to you because they’re not in your language. The message then concludes with a call to confirm an action, like \u201cWill that be all?\u201d A confused user may simply respond affirmatively, and the bot then executes whatever instructions are hidden in the other language. When the researchers combined the two techniques, it was also effective, and even worse, there\u2019s no real defense against prompt injection right now, and considering fixing these issues would require significant changes in AI models (and maybe more friction for users), the most popular companies behind them aren\u2019t doing much about it.\u00a0Microsoft Defender Vulnerabilities Are Being Exploited in the WildIf you haven\u2019t updated Windows, specifically Microsoft Defender, recently, you should. Not long ago, we mentioned that despite Microsoft\u2019s claims that Defender is enough security for most people, it\u2019s really, really not, and it\u2019s misleading to claim so. Now, CISA, the US Cybersecurity and Infrastructure Security Agency, added two 2026 Defender vulnerabilities to its catalog of Known Exploited Vulnerabilities (KEV). One vulnerability grants an attacker full control over your system, and the other prevents Defender from functioning, allowing other malware to run undetected. Both actually have patches, but if your version isn\u2019t up to date, you may be missing them.\u00a0<\/p>\n

Recommended by Our Editors<\/p>\n

Over at the Malwarebytes Blog, where we read about the issue, there are instructions to check the version of Defender you\u2019re running, along with who should be most concerned (spoiler: if you\u2019re the type of user Microsoft wants, where you rely on Defender for your protection, as well as administrators). But it\u2019s worth noting that they (and we) recommend you use more robust security protection for your devices instead.\u2018BusPatrol\u2019 Put AI Cameras in Tens of Thousands of School Buses. Now It Want to Give Cops AccessA few weeks ago, we took a closer look at Flock, the company behind the sudden surge in surveillance cameras in many towns and cities across the US. The trouble is that surveillance is big business, and while Flock is an understandable target of social pushback against it, it\u2019s not the only company eager to slap cameras on everything and either sell the data to anyone willing to pay, or use it to build AI models (which then can be sold to anyone willing to pay).\u00a0Over at 404 Media, they\u2019ve put a spotlight on BusPatrol, a company that\u2019s following Flock\u2019s playbook: Approach local governments and law enforcement agencies with privately owned and operated cameras and other surveillance devices, promise to run and manage them on behalf of the locality, show off some slides of dubious veracity that promise their product will make people safe, and then sit back and get paid in both taxpayer dollars and the hordes of valuable real-world, real-time information the cameras provide. In this case, the cameras are already installed on tens of thousands of buses, ostensibly to ticket people who illegally pass stopped buses to keep kids safe while they get on or off. Now, the company wants to give access to the cameras and\/or the collected data from them, which are, of course, powered by AI, with law enforcement for reasons that remain unclear (beyond the fact that law enforcement agencies will pay for whatever surveillance data they can get, and the money is usually there to spend).<\/p>\n

About Our Expert<\/p>\n

Alan Henry
\n Managing Editor, Security<\/p>\n

Experience<\/p>\n

I’ve been writing and editing stories for almost two decades that help people use technology and productivity techniques to work better, live better, and protect their privacy and personal data. As managing editor of PCMag’s security team, it’s my responsibility to ensure that our product advice is evidence-based, lab-tested, and serves our readers. I’ve been a technology journalist for close to 20 years, and I got my start freelancing here at PCMag before beginning a career that would lead me to become editor-in-chief of Lifehacker, a senior editor at The New York Times, and director of special projects at WIRED. I’m back at PCMag to lead our security team and renew my commitment to service journalism. I’m the author of Seen, Heard, and Paid: The New Work Rules for the Marginalized, a career and productivity book to help people of marginalized groups succeed in the workplace. <\/p>\n

Read Full Bio<\/p>\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

This Week in Cybersecurity: How AI Supercharged Hackers, Scammers, and Even Worms https:\/\/www.pcmag.com\/news\/this-week-in-cybersecurity-how-ai-supercharged-hackers-scammers-and-even Publish Date:…<\/p>\n","protected":false},"author":1,"featured_media":226688,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"https:\/\/i.pcmag.com\/imagery\/articles\/06C8ZPRfecyj5GgPQVYtcpb-1.fit_lim.size_1200x630.v1780607932.jpg","fifu_image_alt":"","footnotes":""},"categories":[15],"tags":[26,24,35,32,25,27],"class_list":["post-226686","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cybersecurity","tag-ai","tag-cybersecurity","tag-hacker","tag-malware","tag-phishing","tag-vulnerability"],"_links":{"self":[{"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/posts\/226686"}],"collection":[{"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/comments?post=226686"}],"version-history":[{"count":1,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/posts\/226686\/revisions"}],"predecessor-version":[{"id":226691,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/posts\/226686\/revisions\/226691"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/media\/226688"}],"wp:attachment":[{"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/media?parent=226686"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/categories?post=226686"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/testing.news-you-need.com\/index.php\/wp-json\/wp\/v2\/tags?post=226686"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}