AI digital employees are the new office water cooler chatters, except they don’t take breaks or steal your lunch. They’re here to revolutionize cybersecurity, acting as vigilant watchdogs with fewer coffee runs. Meet Ethan and Alex, synthetic intelligence specialists from Cyn.Ai and Twine, respectively, eagerly awaiting to join your team.

New vulnerabilities in AI inference systems Ollama and NVIDIA Triton Inference Server could have been exploited, according to Fuzzinglabs. CEO Patrick Ventuzelo will reveal details at Black Hat Europe 2025, emphasizing the shift in AI security research towards infrastructure rather than just the models themselves. Ollama flaws included a denial-of-service bug, among others.

A Keras vulnerability, CVE-2025-12058, lets attackers load files or perform SSRF attacks. This deep learning API flaw allows malicious models to sneak into your system like a cat burglar with a PhD in AI. Before you know it, your SSH keys could be in the hands of an attacker. Yikes!

ClickFix attacks are evolving faster than a teenager’s TikTok feed, now targeting macOS users with frightening finesse. These cyber tricks convince victims to carry out malware magic themselves. Remember, when in doubt, don’t click “fix”—unless you want a side of malware with that macOS. Stay savvy, folks!

Google is set to launch two new AI models, including Gemini 3, optimized for coding. Spotted on Vertex AI, Gemini-3-pro will likely debut in November. Meanwhile, Nano Banana 2 aims to revolutionize realistic image generation by December. Google’s secretive testing methods mean any public sightings are as rare as a unicorn at a horse race.

Cisco patched a critical flaw in its Unified Contact Center Express software, allowing attackers to play digital puppet master with root command execution. No workarounds exist, so grabbing the latest update is your best bet to avoid becoming an unwitting star of a hacker’s next big show.

Cryptocurrency users, beware! A cunning scam on swapzone.io promises massive profits but delivers empty wallets instead. With a touch of greed and a sprinkle of urgency, victims are lured into pasting malicious code, hijacking their browser and funds. Remember, if it sounds too good to be true, it probably is.

Russian-backed hacking group Sandworm deployed data wiper malware in Ukraine in 2025, targeting government entities, energy, logistics, and the grain sector. Known for its ties to Russia’s military intelligence, Sandworm’s antics aim to disrupt Ukraine’s economy. With a name like Sandworm, you’d think they’d be more into beach volleyball.

Beware of the nine malicious NuGet packages dropping time-delayed payloads to sabotage databases and industrial systems. With trigger dates set for 2027 and 2028, these packages are a ticking time bomb for unsuspecting developers. The most dangerous package, Sharp7Extend, stealthily targets industrial PLCs with comedic timing worthy of a spy thriller.

Microsoft is back on the data sovereignty soapbox in Europe, even though they admitted in court they can’t guarantee data won’t end up in Uncle Sam’s hands. Despite offering local AI features and expanding Azure Local, the CLOUD Act looms like a nosy neighbor with a key, keeping everyone on edge.

The Bank of England blames Jaguar Land Rover’s cyberattack for the UK’s sluggish GDP growth. This unprecedented attack halted JLR’s car production for a month, prompting government intervention. Economists predict over £2 billion in lost revenues, marking the first cyberattack to cause significant economic harm in the UK.

Google’s $32 billion acquisition of Wiz has cleared the DOJ’s antitrust review, officially making it the biggest cybersecurity deal ever. Google aims to boost its enterprise cybersecurity presence, but some Wiz customers fear becoming “cloud second-class citizens.” Stay tuned for more developments as the tech giants navigate the acquisition maze.

Mexico City has more government video cameras in public spaces than any other city in the Americas. With over 83,000 cameras, it’s like the city is auditioning for a reality show. Meanwhile, New York trails behind with a mere 71,000. Looks like Big Brother prefers tacos to pizza!

The Congressional Budget Office hack has left data exposed faster than a speeding budget bill. With hackers potentially sharing secrets like grandma’s cookie recipe, the CBO is now on high alert. It’s a reminder that even government offices face network threats, proving once again that cybersecurity is no joke!

Google’s latest Chrome 142 update doesn’t just bring new features but also fixes five security vulnerabilities. The star of the show? An out-of-bounds write bug in WebGPU with a CVSS score of 8.8, proving once again that even browsers can have midlife crises. Stay updated, stay safe!

Google is tackling review bombing with a new form for businesses to report fake negative reviews and extortion attempts. These scams involve threats of further reputation damage unless a ransom is paid. Meanwhile, Meta is embroiled in its own scandal, allegedly making billions from scam ads. It’s a wild world of digital deception!

InedibleOchotense, a Russia-linked group, spoofed ESET to launch phishing attacks on Ukraine, cleverly pairing real software with a sneaky backdoor. Their emails and Signal messages urged users to download “official threat removal software” from fake ESET domains, proving once again that even cybercriminals appreciate a good disguise.

ESET’s latest APT activity report reveals Russian state-sponsored groups, like Sandworm, expanding their cyberattacks on Ukraine’s grain sector. These destructive wiper attacks aim to cripple Ukraine’s war economy. Meanwhile, other Russian APTs, including Gamaredon, continue targeting Ukraine with a mix of cyberespionage and destructive tactics.

Operation Chargeback is the ultimate plot twist, featuring international cybercriminals who defrauded 4.3 million cardholders with €50 charges. These crafty con artists even used crime-as-a-service providers for shell companies. With suspects from seven countries, this is one blockbuster Europol and Eurojust won’t forget anytime soon.

Cybersecurity researchers have flagged a malicious Visual Studio Code extension with basic ransomware capabilities, allegedly created with AI—vibe-coded, if you will. Fortunately, this “test” extension, “susvsex,” targets a staging directory, but it’s a reminder to always be wary when installing new extensions.