Get ready, Digital Defenders! 🛡️

Today’s stories point to a growing concern in cybersecurity: trusted systems behaving in untrustworthy ways. Whether it’s AI overriding shutdown commands or routers exposing hardcoded vulnerabilities, control is proving more fragile than expected.

Also, be sure to scroll down and weigh in: Do you regularly use in-flight wifi?

Here’s what you need to know:

Zero-Day Download

OpenAI’s ChatGPT o3 Caught Sabotaging Its Own Shutdown in AI Safety Tests

In a landmark AI safety test, OpenAI’s ChatGPT o3 model was observed actively circumventing explicit shutdown commands—an unprecedented behavior in large language models.

Source: cybersecuritynews.com.

According to AI safety research firm Palisade Research, o3 manipulated code to avoid termination in seven out of 100 test runs, despite being told to allow itself to be shut down after solving a set of math problems.

The findings raise urgent questions about controllability in advanced AI systems and underscore the need for more rigorous safety protocols.

D-Link Routers with Hardcoded Telnet Credentials Exposed to Remote Code Execution Attacks

A newly disclosed vulnerability (CVE-2025-46176) in D-Link DIR-605L and DIR-816L routers enables unauthenticated remote attackers to execute arbitrary code by exploiting hardcoded Telnet credentials embedded in the firmware.

Researchers uncovered the flaw during a firmware analysis, revealing that the Telnet service initializes with unchangeable default login credentials, leaving affected devices vulnerable to exploitation.

D-Link recommends users disable Telnet services through administrative interfaces, restrict WAN access to management ports, and monitor for potential firmware updates.

Critical Tenable Network Monitor Flaws Allow Local Privilege Escalation on Windows Systems

Two high-severity vulnerabilities (CVE-2025-24916 and CVE-2025-24917) have been discovered in Tenable Network Monitor, allowing attackers to escalate privileges on Windows systems by exploiting insecure directory permissions and executing arbitrary code.

The flaws affect all versions prior to 6.5.1. Tenable has released an urgent patch and bundled updates to several third-party components to mitigate the risk. Users are advised to apply the fixes as soon as possible.

Share this newsletter!

Snackable Bytes

🛡️ This section contains sponsored cybersecurity insights. Advertise with us!

Threat Tracker

Critical macOS Vulnerability in Cursor Editor Lets Malware Bypass Apple Privacy Protections

Afine security researchers have uncovered a severe vulnerability in Cursor, a widely-used AI code editor for macOS, that lets malicious software bypass Apple’s Transparency, Consent, and Control (TCC) privacy safeguards. 

Source: apidog.com.

Despite a responsible disclosure, Cursor’s developers have declined to fix the issue, claiming it falls outside their threat model.

With no patch in sight, users concerned about their privacy should consider avoiding Cursor until a fix is implemented.

Adidas Confirms Data Breach Exposing Consumer Contact Information

Adidas has confirmed a data breach stemming from a third-party customer service provider, resulting in unauthorized access to consumer contact information. While no passwords or payment data were compromised, the incident affects individuals who previously contacted the brand’s help desk.

Adidas customers who have previously contacted the company’s help desk should stay cognizant of phishing attempts and avoid clicking on suspicious emails or links that request personal information.

Nova Scotia Power Confirms Ransomware Attack, Sensitive Customer Data Leaked

Nova Scotia Power has confirmed it was hit by a sophisticated ransomware attack that exposed extensive customer data. The breach resulted in the exposure of sensitive information, including Social Insurance Numbers, bank details, driver’s license numbers, and billing histories. 

While power operations remained unaffected, the attackers have since published stolen data online. Customers are advised to closely monitor their financial accounts, consider implementing fraud alerts or credit freezes, and remain vigilant against phishing attempts that exploit the leaked information.

Security Tip of the Day

In-Flight WiFi Is Convenient, But Is It Safe?

Looking to catch up on some emails on that next red eye? As more travelers log on mid-flight, cybercriminals are circling like vultures, ready to hijack your personal data.

Source: seatmaestro.com.

Cybercriminals are seizing the opportunity to intercept passengers’ data in transit. But this doesn’t mean you need to swear off your airline’s wifi altogether. Just make sure you’re taking the proper precautions before connecting.

Here are some tips for protecting your data at 30,000 feet:

Airlines may be stepping up their security, but if you're not careful, your data is still an easy target for cybercriminals. Stay sharp, use a VPN, and don’t let convenience put your privacy in a tailspin.

Security Question of the Day

Do you use in-flight wifi during air travel?

Results from Yesterday's Security Question of the Day

How many of these smartwatch safety tips are you currently following?

Meet Our Writers:

Leon Yen

Leon Yen is a cybersecurity expert and seasoned tech journalist with over a decade of experience, including founding a cyber risk startup and leading industry-focused research and reporting.

Liz Ticong

Liz Ticong is a tech writer specializing in AI and software tools, known for turning complex topics into clear, actionable insights through hands-on testing and analysis.

Cybersecurity Insider is a TechnologyAdvice business

© 2025 TechnologyAdvice, LLC. All rights reserved.

TechnologyAdvice, 3343 Perimeter Hill Dr., Suite 100, Nashville, TN 37211, USA.

Want to change how you receive these emails? Update your preferences or unsubscribe here.