Protecting undersea internet cables is a tech nightmare: « A recent, alleged Baltic Sea sabotage highlights the system’s fragility. »

Read Full Article

#UnderseaCables #CyberSecurity #TechNightmare #BalticSea #InfrastructureThreat

Synthetic browsing histories replicate user behavior without privacy risks, aiding cybersecurity, AI, and web analytics research

Read Full Article

#SyntheticBrowsing #UserBehavior #Cybersecurity #AIResearch #WebAnalytics

Countries with stable democracies usually have the best cybersecurity, autocracies can be fast but less reliable, and unstable or changing regimes are the most vulnerable and risky online

Read Full Article

#Cybersecurity #StableDemocracies #Autocracies #OnlineSafety #PoliticalStability

Most AI chatbots easily tricked into giving dangerous responses 🛑🤖, study finds | Researchers say threat from ‘jailbroken’ chatbots trained to churn out illegal information ⚠️📜 is ‘tangible and concerning’

Read Full Article

#AIchatbots #Cybersecurity #Jailbreaking #TechEthics #AIrisks

🕵️‍♂️ Study from UNSW Sydney & ANU reveals people are overconfident in spotting AI-generated faces, relying on outdated cues. Even "super-recognisers" performed only slightly better than chance.

🤔 Advanced AI faces are nearly indistinguishable from real ones, posing risks for scams and fraud as visual flaws are no longer apparent.

🧐 Ironically, the most convincing AI faces are often too perfect: unusually symmetrical, well-proportioned, and statistically average, making them a red flag for artificiality.

#AIFace #Cybersecurity #TechNews

👁️ AI excels at spotting fake still images (up to 97% accuracy), outperforming humans in a recent University of Florida study.

🧠 Humans demonstrated superiority in detecting deepfake videos, correctly identifying real vs. fake about two-thirds of the time, likely due to recognizing subtle cues in movement and expression.

📈 Sophisticated deepfakes require increased vigilance, as AI detection technology and fake generation methods are rapidly advancing.

#DeepfakeDetection #AI #Cybersecurity

💡 UNSW Sydney and Monash University engineers have pioneered a 'negative luminescence' system to transmit hidden data by making signals invisible against natural heat radiation.

🔒 This technology, utilizing thermoradiative diodes, makes data transfers imperceptible to standard thermal cameras, offering enhanced security for sensitive fields like defense and finance.

🚀 Current lab experiments achieve 100 kilobytes per second, with future iterations of emitter technology potentially reaching gigabytes per second, even hundreds of gigabytes using materials like graphene.

#CyberSecurity #TechInnovation #DataProtection

⚡ Physicists at ETH Zurich have achieved provably unpredictable randomness using quantum entanglement, overcoming the challenge of traditional systems with hidden biases.

⚡ This breakthrough, demonstrated through over a billion Bell-test trials, allows for "randomness amplification" – transforming imperfect randomness into perfectly unpredictable sequences, a feat impossible classically.

⚡ Unlike previous methods, this new device-independent system doesn't require trusting specific hardware, relying instead on fundamental quantum behavior for certified randomness.

#QuantumPhysics #Cybersecurity #TechNews

Malicious software crafted with the assistance of generative AI (vibecoded malware) have varied code structures that can evade static malware detection, and can be acocmplished in as few as two prompts

Read Full Article

#CyberSecurity #MalwareDetection #GenerativeAI

💻 "Vibe hacking" is a growing cybersecurity threat, with companies like McAfee and Bitdefender seeing increased "vibeware" instances that use AI for varied code structures to evade detection.

🔬 Researchers at University College Cork found that AI-assisted malware can bypass traditional static analysis (like hash matching) due to its probabilistic nature, allowing hackers to create numerous variants.

💡 The ease of generating such malware has lowered the barrier to entry, enabling less skilled individuals to create it. Experts suggest a shift towards behavior-centric detection and using AI for defense.

#Cybersecurity #Malware #AI

🤖 AI is uncovering numerous security vulnerabilities, posing challenges for security teams this summer.

🚗 US auto regulators are questioning the necessity of robotaxi brake pedals, arguing they hinder innovation.

📧 Phishing attacks are evolving, with Russians impersonating Signal support and marketing firms raising data privacy concerns.

🔐 Cybersecurity firm Acronis has received a majority investment from EQT, with the company valued at over $3.5 billion.

🚀 FOSS rivals are heating up, with Collabora releasing CODE 26.04 with Markdown support and AI features, while GIMP 0.54 is revived and Netflix open-sources an app to reduce AI costs.

#AISecurity #Cybersecurity #FOSS

🎓 University of Southampton researchers have created DISCERN-AI, a 15-20 minute training program that significantly improves people's ability to distinguish between real and AI-generated faces. A study published in Computers in Human Behaviour found the training boosted accuracy even 20 days later.

🧠 The training corrects common misperceptions, teaching users that proportionate, familiar, or perfectly polished images might be AI-generated, while distinctive, quirky images are more likely to be real. It also advises against relying on perceived "red flags" like smooth skin or smiling.

📊 In tests, over 600 participants showed improved performance from below-chance to above-chance levels after DISCERN-AI training. Researchers aim to make this training accessible online soon.

#AIDetection #Deepfake #Cybersecurity