TechLetters ☕️Humanitarian concerns of cyber operations and misinformation. iPhones automatically reboot after 72h. Should humanity give human rights to Large Language Models?
Security
Humanitarian call for curbs on cyberattacks and information operations. Strong message from International Conference of the Red Cross and Red Crescent warning about cyberattacks and misinformation. Also a note to private (cyber/tech/IT) businesses: "providing services to clients that are or may become involved in armed conflict involves certain risks" [including to those businesses or their employees] But also notable: amplify harm to the civilian population and objects, including when communication tech is used to recruit children into armed forces., and this isn’t being spoken in public that much Another very relevant concern: cyber techs may enable or be used to encourage civilians to conduct or support activities in armed conflict, while civilians may not be aware of the risks involved or the implications applicable to their conduct. Disinformation and propaganda becomes a significant concern for humanitarian operations. I tackle the issue in my book as well.
NIST standardisation organisation says that systems must phase out non-quantum-resistant cryptography by 2035. RSA, ECDSA, ECDH disallowed. https://nvlpubs.nist.gov/nistpubs/ir/2024/NIST.IR.8547.ipd.pdf
United Nations Security Council discussed cyberattacks on healthcare systems. US, China, Russia, France are all concerned. While no cyberattacks aimed to cause direct fatalities, they threaten healthcare operations and could indirectly increase mortality. I consider these points at length in “Philosophy of Cybersecurity”.
iPhone left without being unlocked reboots itself after 72 hours. After few days, the device automatically restart. For security reasons. This is an implementation of National Security Agency recommendation to protect from being hacked and data theft. Inactivity reboot makes the device more secure (reduces risk of hacking/compromise, data theft, cybercriminal success, or petty thiefs) and is in line with U.S. NSA security recommendations. Technical details here. “While the media coverage so far framed this mitigation as primarily targeting law enforcement, it also a huge security improvement against theft. Outdated law enforcement equipment often finds its way to eBay and other similar platforms for rather cheap price tags. However, thieves won't have the financial and legal means to obtain up-to-date exploits to unlock iPhones within 3 days of getting them.”
Privacy
Smart TVs watching you. Smart TVs use Automatic Content Recognition (ACR) to track what’s on screen—shows, ads, movies—and build user profiles. Opting out can stop ACR data traffic, though privacy settings are often hidden.
Consent form to remove third-party cookies from Chrome unclear. Google hasn’t decided how it should look like just yet. But it considers that cookie less traffic will be growing, and “Privacy Sandbox remains critical for businesses”.
Other
Qwen-2.5-Coder 32B LLM model offers GPT4-grade model on a laptop. Well, the strong ones with lots of RAM. That's the best privacy-preserving LLM model out there: running on the device.
Anthropic appointed an “AI welfare” expert. To assess whether its systems are inching towards consciousness or agency, and whether their welfare must be considered. Should we seriously give human rights to large language models? Are prompts moral?
In case you feel it's worth it to forward this content further:
Subscribed
If you’d like to share: