Chinese AI models have matched the performance of Anthropic’s powerful model Mythos in some cybersecurity scenarios.
Researchers point to Zhipu AI's open-weight GLM-5.2 and a new tool, Tulongfeng, from 360 Security Technology, whose CEO said the capability cannot stay "solely in American hands."
The Trump administration has kept the more capable Anthropic models, Fable and Mythos, off-limits to foreign users for more than two weeks while continuing to clear AI chip exports to China, a pairing one former export-control official called "a gift to China."
https://www.internationalcyberdigest.com/china-matches-mythos-on-cyber-bug-finding-as-the-us-restricts-its-own-model/
Researchers point to Zhipu AI's open-weight GLM-5.2 and a new tool, Tulongfeng, from 360 Security Technology, whose CEO said the capability cannot stay "solely in American hands."
The Trump administration has kept the more capable Anthropic models, Fable and Mythos, off-limits to foreign users for more than two weeks while continuing to clear AI chip exports to China, a pairing one former export-control official called "a gift to China."
https://www.internationalcyberdigest.com/china-matches-mythos-on-cyber-bug-finding-as-the-us-restricts-its-own-model/
❤9🔥3🥰2🤣1
This media is not supported in your browser
VIEW IN TELEGRAM
In the Russian town of Podolsk, a 13-year-old boy set fire to a fuel pump at a gas station. According to Russia's Interior Ministry, online handlers had tricked him into doing it.
The boy had been chatting with someone online he thought was a girl, and he'd shared his location with "her." The handlers then told him that location was going to be used to target a Ukrainian missile strike, and that the only way to stop the strike was to burn down the gas station.
The boy had been chatting with someone online he thought was a girl, and he'd shared his location with "her." The handlers then told him that location was going to be used to target a Ukrainian missile strike, and that the only way to stop the strike was to burn down the gas station.
🤣19😭8❤4👍1🤬1
‼️ iPhone 18 Pro photos were LEAKED as a result of the Tata Electronics breach we reported about recently.
On the specs: the A20 Pro is reportedly fabbed by TSMC on its first 2nm (N2) process, paired with 96-bit-wide LPDDR6 and a reported 12GB on the 18 Pro.
Tata builds about a third of Apple's iPhones in India, more than 204,000 documents totalling 630+ GB have been leaked including Tesla and Apple confidential documents.
https://www.internationalcyberdigest.com/tata-electronics-leaked-iphone-18-pro-photos-and-designs/
On the specs: the A20 Pro is reportedly fabbed by TSMC on its first 2nm (N2) process, paired with 96-bit-wide LPDDR6 and a reported 12GB on the 18 Pro.
Tata builds about a third of Apple's iPhones in India, more than 204,000 documents totalling 630+ GB have been leaked including Tesla and Apple confidential documents.
https://www.internationalcyberdigest.com/tata-electronics-leaked-iphone-18-pro-photos-and-designs/
❤10💩1
‼️ iPhone 18 Pro photos were LEAKED as a result of the Tata Electronics breach we reported about recently.
On the specs: the A20 Pro is reportedly fabbed by TSMC on its first 2nm (N2) process, paired with 96-bit-wide LPDDR6 and a reported 12GB on the 18 Pro.
Tata builds about a third of Apple's iPhones in India, more than 204,000 documents totalling 630+ GB have been leaked including Tesla and Apple confidential documents.
https://www.internationalcyberdigest.com/tata-electronics-leaked-iphone-18-pro-photos-and-designs/
On the specs: the A20 Pro is reportedly fabbed by TSMC on its first 2nm (N2) process, paired with 96-bit-wide LPDDR6 and a reported 12GB on the 18 Pro.
Tata builds about a third of Apple's iPhones in India, more than 204,000 documents totalling 630+ GB have been leaked including Tesla and Apple confidential documents.
https://www.internationalcyberdigest.com/tata-electronics-leaked-iphone-18-pro-photos-and-designs/
💩9🤣5❤1
‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message.
Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice.
A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.
Read:
https://www.internationalcyberdigest.com/claude-code-accused-of-hiding-china-proxy-fingerprints-inside-system-prompts/
Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice.
A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.
Read:
https://www.internationalcyberdigest.com/claude-code-accused-of-hiding-china-proxy-fingerprints-inside-system-prompts/
💩15❤7🔥3🤣3👏2👍1
‼️Researchers found six wireless pre-authentication vulnerabilities in Apple AirDrop and Google/Samsung Quick Share, the proximity file-transfer protocols running on over 5 billion devices. The bugs are reachable from wireless range with no pairing.
Five are denial-of-service or protocol-manipulation flaws. One, a use-after-free in Quick Share for Windows, earned a Google bounty and is rated potentially exploitable. The team says it could not bypass the file-transfer consent prompt.
Apple, Samsung, and Google have acknowledged the reports Most are not yet patched, and no CVEs have been assigned, with one pending for the Windows bug.
Read: https://www.internationalcyberdigest.com/researchers-find-six-new-vulnerabilities-in-airdrop-and-quick-share-affecting-5-billion-devices/
Five are denial-of-service or protocol-manipulation flaws. One, a use-after-free in Quick Share for Windows, earned a Google bounty and is rated potentially exploitable. The team says it could not bypass the file-transfer consent prompt.
Apple, Samsung, and Google have acknowledged the reports Most are not yet patched, and no CVEs have been assigned, with one pending for the Windows bug.
Read: https://www.internationalcyberdigest.com/researchers-find-six-new-vulnerabilities-in-airdrop-and-quick-share-affecting-5-billion-devices/
❤2
‼️ Huntress' CEO says it's not illegal that an employee of theirs tipped off a threat actor about the FBI looking for them, then closes his blog post with "protecting ALL businesses while wrecking adversaries in the process."
How exactly are you wrecking adversaries when you act as their informant by allegedly sending them screenshots that named FBI agents?
Their CEO, Kyle Hanslovan, said in an interview today that a current employee told a ransomware actor that law enforcement had reached out asking about that actor. He calls it "poor judgment," says it was not illegal, and rejects the "insider threat" label.
The former employee who raised it, Ben Folland, says that admission proves his point. He alleges the employee forwarded FBI communications to the DevMan ransomware group, then refused to cooperate with the FBI. His supporting evidence is not public yet.
Threat-intel teams routinely talk to criminals for research, but once a researcher warns a target that law enforcement is circling, an active case can collapse, and that is the line Huntress now has to defend in public.
We've reached out to Huntress with questions about the employee who allegedly sent DevMan information, and they pointed us to their CEO's blog post.
Article: https://www.internationalcyberdigest.com/huntress-admits-an-employee-tipped-a-ransomware-actor-to-law-enforcement/
How exactly are you wrecking adversaries when you act as their informant by allegedly sending them screenshots that named FBI agents?
Their CEO, Kyle Hanslovan, said in an interview today that a current employee told a ransomware actor that law enforcement had reached out asking about that actor. He calls it "poor judgment," says it was not illegal, and rejects the "insider threat" label.
The former employee who raised it, Ben Folland, says that admission proves his point. He alleges the employee forwarded FBI communications to the DevMan ransomware group, then refused to cooperate with the FBI. His supporting evidence is not public yet.
Threat-intel teams routinely talk to criminals for research, but once a researcher warns a target that law enforcement is circling, an active case can collapse, and that is the line Huntress now has to defend in public.
We've reached out to Huntress with questions about the employee who allegedly sent DevMan information, and they pointed us to their CEO's blog post.
Article: https://www.internationalcyberdigest.com/huntress-admits-an-employee-tipped-a-ransomware-actor-to-law-enforcement/
💩9❤6👍1😁1