It was only a matter of time before hackers started using artificial intelligence to attack artificial intelligence—and now that time has arrived. A new research breakthrough has made AI prompt injection attacks faster, easier, and scarily effective, even against supposedly secure systems like Google’s Gemini. Prompt injection attacks have been one of the most reliable ways to manipulate large language models (LLMs). By sneaking malicious instructions into the text AI reads—like a comment in a block of code or hidden text on a webpage—attackers can get the model to ignore its original rules. That could mean leaking private data, delivering wrong answers, or carrying out other unintended behaviors. The catch, though, is that prompt injection attacks typically require a lot of manual trial and error to get right, especially for closed-weight models like GPT-4 or Gemini, where developers can’t see the underlying code or training data. But a new technique called Fun-Tuning changes that. Developed by a team of university researchers, this method uses Google’s own fine-tuning API for Gemini to craft high-success-rate prompt injections—automatically. The researcher’s findings are currently available in a preprint report. By abusing Gemini’s training interface, Fun-Tuning figures out the best “prefixes” and “suffixes” to wrap around an attacker’s malicious prompt, dramatically increasing the chances that it’ll be followed. And the results speak for themselves. In testing, Fun-Tuning achieved up to 82 percent success rates on some Gemini models, compared to under 30 percent with traditional attacks. It works by exploiting subtle clues in the fine-tuning process—like how the model reacts to training errors—and turning them into feedback that sharpens the attack. Think of it as an AI-guided missile system for prompt injection. Even more troubling, attacks developed for one version of Gemini transferred easily to others. This means a single attacker could potentially develop one successful prompt and deploy it across multiple platforms. And since Google offers this fine-tuning API for free, the cost of mounting such an attack is as low as $10 in compute time. Google has acknowledged the threat but hasn’t commented on whether it plans to change its fine-tuning features. The researchers behind Fun-Tuning warn that defending against this kind of attack isn’t simple—removing key data from the training process would make the tool less useful for developers. But leaving it in makes it easier for attackers to exploit. One thing is certain, though. AI prompt injection attacks like this are a sign that the game has entered a new phase—where AI isn’t just the target, but also the weapon. Don't Miss: Lose weight while eating junk food? Scientists may have found a way The post Hackers are now using AI to break AI – and it’s working appeared first on BGR. Today's Top Deals Today’s deals: $50 off new M4 MacBook Air, $57 Insignia smart TV, $199 DJI Osmo action camera, more Amazon Big Spring Sale: Apple deals, gaming laptops, Crest 3D Whitestrips, Samsung deals, more Amazon Big Spring Sale: $299 Apple Watch S10, $350 70″ smart TV, $299 King mattress, $58 myQ, more Today’s deals: $169 AirPods Pro 2, $100 off Breville air fryers, $40 Echo Dot, $250 Sony XM5 headphones, more

Microsoft has officially declared war on the password. In a sweeping update affecting more than a billion users, the company is making it clear—it’s time to ditch your Microsoft account password for good. This is just the latest move in Microsoft's passkey update, which aims to move all users away from the security wyas of olden days. Starting in April, Microsoft will begin rolling out a new sign-in and account creation experience that puts passkeys at the center. “Our ultimate goal is to remove passwords completely,” the company said in a security update posted in December. Microsoft says it now blocks around 7,000 password-related attacks per second, nearly double the rate from last year. With AI-fueled phishing attempts and increasingly clever hacks, passwords—no matter how long or quirky—just aren’t holding up. Forcing a passkey on Microsoft users seems to be the easiest way to address the problem. That’s where the passkey comes in. This credential is tied to your physical device and unlocked by something only you have—like a fingerprint, face scan, or device PIN. Unlike a password, a passkey can’t be phished, guessed, or intercepted. It’s stored securely on your device and never leaves it. Password managers like Proton Pass and 1Password will let you save passkeys for internet accounts. Image source: 1Password More importantly, it’s fast. Microsoft says passkeys are not only more secure but three times faster than typing in a traditional password. And the transition is already underway. When creating a new Microsoft account, you won’t be asked to set a password. Instead, you’ll verify your email once and then create a passkey. For existing accounts, the sign-in experience is being redesigned to push passkeys as the default to nudge users toward a truly passwordless future. That’s because having a passkey isn’t enough if you’re still keeping the old password around “just in case.” According to Microsoft, that’s like locking your front door but leaving the window wide open for anyone to enter. The presence of a password—even as a backup—leaves your account open to phishing, brute-force attacks, and social engineering scams. That’s why the company says this isn’t just a shift in preference. Microsoft's passkey update is a massive security imperative. Millions of users have already deleted their passwords, according to Microsoft. And this change is about scaling that momentum across its entire user base. Microsoft’s bold move sets a new bar—but not everyone is sprinting toward it. Google, for instance, still supports passwords as fallback credentials, which keeps that potential vulnerability alive. Security researchers and privacy advocates argue that consistency across platforms will be key to making passwordless systems mainstream. For now, Microsoft is leading the charge, both in tech and in messaging clarity. Don't Miss: This AI can tell your biological age just by looking at a few drops of blood The post Microsoft wants you to delete your password and no, it’s not a gimmick appeared first on BGR. Today's Top Deals Today’s deals: $269 iPad 10, $90 Ninja blender, 20% off Coop pillows, $400 Samsung M8 monitor, more Amazon gift card deals, offers & coupons 2025: Get $2,000+ free Best deals: Tech, laptops, TVs, and more sales Today’s deals: $50 off new M4 MacBook Air, $57 Insignia smart TV, $199 DJI Osmo action camera, more

Google’s newest budget phone is finally ready for release. After an inexplicable delay due to a “component quality issue,” the company has confirmed the Pixel 9a release date: April 10 in the US, Canada, and the UK. The rest of the world will follow shortly after, with launches in Europe on April 14 and select Asian-Pacific regions on April 16. We’ve known about the Pixel 9a for at least a week now—with official confirmation giving us a price and a good look at the design. The Pixel 9a might look like a flagship at first glance, sharing the same sleek design language and housing Google’s Tensor G4 chip. But look closer, and it’s clear that a few of Google’s headline features didn’t make the trip down to this more affordable device. That’s especially true when it comes to AI. Despite sporting the same chip as the Pixel 9 and 9 Pro, the 9a has a tighter memory ceiling at just 8GB of RAM. That limitation means the phone can’t run the full version of Gemini Nano, Google’s powerful on-device AI model. Instead, it comes with a lightweight, text-only variant. Don't expect the same level of Gemini support on the 9a as Google's flagship devices. Image source: Christian de Looper for BGR In practical terms, that means you won’t be using the 9a for AI-powered voice summarization or multimodal features like image-based Q&A and contextual suggestions, which are available on Google’s flagship devices. This is likely an intentional move by Google, aimed at balancing affordability with performance. But it also means buyers hoping for a full suite of on-device AI tools will need to temper their expectations. That’s not the only trade-off. The Pixel 9a also skips satellite communication support and features an older cellular modem, which could affect signal efficiency and battery life in fringe areas. Still, for $499 in the US (with a $100 bump for double the storage), the 9a offers a competitive entry point into the Pixel ecosystem. Pricing varies slightly by region, with a price tag of $679 CAD in Canada, £499 in the UK, and €549 in most of Europe. Interestingly, Japan—a regular participant in Google’s device launches—still lacks a confirmed Pixel 9a release date, though Google insists it’s coming “soon.” If you’ve already signed up for availability notifications through the Google Store, you’ll be among the first to know when preorders go live. And if you’re after a clean Android experience with a side of pared-down AI, the 9a could still hit the sweet spot. Just don’t expect the same Gemini-powered goodies its flagship siblings enjoy. Don't Miss: Google Maps will be able to scan your screenshots to help you plan a trip The post Pixel 9a will arrive on April 10 missing some of Google’s best AI features appeared first on BGR. Today's Top Deals Amazon Big Spring Sale: $269 iPad 10, Nintendo Switch games, robot vacuum deals, $390 AR glasses, more Spring Sale: The ultimate Reolink Wireless Security Camera System is up to 40% off Amazon gift card deals, offers & coupons 2025: Get $2,000+ free Amazon Big Spring Sale: Apple deals, gaming laptops, Crest 3D Whitestrips, Samsung deals, more

ChatGPT's 4o image generation model is the talk of the town right now, but it's not the only AI software that can offer mind-blowing image generation. TikTok parent company ByteDance has a new AI model called InfiniteYou, whose sole purpose is to let users generate photos of themselves starting from a single uploaded photo. It's not that ChatGPT's new image generation powers can't edit photos you upload to the chatbot while preserving the identity of those characters. Other AI tools exist to let you edit your images in ways that fit your needs, even if that essentially means creating fakes; photos showing events that never happened and people who weren't in that picture when it was taken. However, the purpose of ByteDance's new model is to generate fake pictures of a real subject while preserving their identity. That's the whole point of InfiniteYou: To let you create any sort of image, starting from a simple photo upload that contains the main subject and a text prompt that describes what you want the AI to generate. Continue reading... The post ByteDance’s InfiniteYou AI lets you create infinite fake photos of yourself appeared first on BGR. Today's Top Deals Mother’s Day gift ideas 2024: Thoughtful gifts mom will never forget Spring Sale: The ultimate Reolink Wireless Security Camera System is up to 40% off Here are our favorite Magenta Status perks you can only get from T-Mobile Amazon gift card deals, offers & coupons 2025: Get $2,000+ free

The new ChatGPT 4o image generation model is the talk of the town, and not just for good reasons. Everyone is marveling at the AI's amazing new abilities, which include generating legible text in images, creating fake photos out of real ones, creating deepfakes of celebrities, and replicating copyrighted content like Studio Ghibli characters. It all happens incredibly fast, with the AI able to respond to your needs. But some people have been quick to point out the bad things about the new AI image model. First, the most obvious problem that we're not really talking about is that ChatGPT has dealt a swift blow to all sorts of content creators, including graphic designers and photographers. Of course, we already have other AI image-generation programs that endanger those professions. This isn't a ChatGPT safety issue, either. The fact that ChatGPT-created images have no visible watermark to inform unsuspecting people they're not real images is a big safety concern. More visible is the Studio Ghibli controversy, which shows that OpenAI is willing to let 4o image generation easily rip off copyrighted content. The even more annoying thing about ChatGPT's new image generation abilities is how easy it is to make deepfakes of celebrities. This one is especially troubling to me, an internet user, because malicious actors have unfettered access to the tool. OpenAI has started paying attention to the criticism it received since the launch of 4o image generation, but it's not taking any action, especially on the deepfake problem. It turns out the only way to stop someone from using your face with ChatGPT is to opt out of it with OpenAI. Continue reading... The post ChatGPT celebrity deepfakes are going viral, and there’s only one way to stop them appeared first on BGR. Today's Top Deals Today’s deals: $269 iPad 10, $90 Ninja blender, 20% off Coop pillows, $400 Samsung M8 monitor, more Today’s deals: $160 iPhone SE 3, 35% off Red Bull, Sonos speakers, 48% off DEWALT power tools, more Today’s deals: $99 Beats Pill Bluetooth speaker, 20% off LG C3 OLED TV, free Amazon Music streaming, more Amazon gift card deals, offers & coupons 2025: Get $2,000+ free