Two prominent Ethereum figures are sounding the alarm on AI's potential to disrupt cryptocurrency security. Justin Drake from the Ethereum Foundation warned that AI could break ECDSA, the signature system securing Bitcoin and Ethereum wallets, within months in a worst-case scenario. Vitalik Buterin echoed the concern, predicting AI breakthroughs could weaken lattice-based cryptography and ECDSA within two years. Both recommended moving funds to addresses that have never sent a transaction and preparing carefully rather than panicking.
Meanwhile, the line between AI and superintelligence is getting blurred in public discourse. US President Donald Trump has tried renaming artificial intelligence as "super intelligence" to make it sound more impressive. Philosopher Nick Bostrom clarifies that superintelligence means an intellect beating all humans in every cognitive field, a threshold current narrow or general AI systems have not reached.
On the image generation front, Google Gemini produced a realistic-looking AI fake photo of a FedEx Office, but subtle errors surfaced, such as a lanyard disconnected from a badge and a nonsensical slogan. Experts note AI still struggles with perspective, small text, and object relationships, even as generation quality keeps improving.
In consumer safety, AI chatbots and fake online sellers are creating new risks. A test showed ChatGPT eventually recommended unsafe newborn sleep products despite initial safety guidance. AI-generated social media videos featuring fabricated people are also being used to sell products, exposing gaps in traditional product safety enforcement since identifying anonymous AI-generated sellers remains difficult.
SK Telecom and Upstage launched the Sovereign AI Cybersecurity Consortium with six security firms and three universities in South Korea. The group will use SK Telecom's A.X K series and Upstage's Solar models trained on real-world security data, tested through zero-day vulnerability analysis and real-time monitoring, with mutual red-teaming and university evaluation supporting the effort.
Tom Campbell, CEO of E-ACT, was named Chair of Oak National Academy, where he will oversee the digital national curriculum and integration of AI tools including Aila. The appointment signals a new phase for AI work in education, with Campbell guiding innovation alongside the existing Oak National Academy team.
AI is also reshaping how parents view creative college degrees. One parent shared growing worry about their children's majors in music production and creative writing, as AI's rise has altered perceptions of future job prospects in creative fields.
In a related trust issue, experts examine the concept of reliable narrators in the AI era. With AI systems now generating text, images, and videos, concerns about truth and accuracy have intensified. Proposed solutions include using AI to support human judgment rather than replace it, and building more transparent and explainable AI systems.
AI-powered drones are aiming to dramatically speed up search and rescue missions. DEEMI is partnering with the University of Maine and the University of Massachusetts Amherst to develop drones that flag images matching a missing person's description, potentially cutting mission time from 72 hours down to 7 minutes. The AI is still being trained and is expected to be operational by year-end.
Hurricane forecasting is also benefiting from AI. During Hurricane Isaias, the National Hurricane Center leaned on Google AI models alongside traditional methods. Google DeepMind's model outperformed both American and European traditional models during the 2025 hurricane season, and the Center expects AI models to become increasingly influential in future forecasting.
Key Takeaways
- Justin Drake warned AI could break ECDSA wallet security within months; Vitalik Buterin gave a two-year timeline for AI weakening lattice-based cryptography and ECDSA.
- Both Ethereum figures recommend moving funds to addresses that have never sent a transaction and preparing carefully rather than panicking.
- Donald Trump has tried renaming artificial intelligence as superintelligence, though experts like Nick Bostrom say current AI is still narrow or general, not true superintelligence.
- Google Gemini generated a realistic but flawed AI fake photo of a FedEx Office, revealing AI still struggles with perspective, small text, and object relationships.
- A test showed ChatGPT eventually recommended unsafe newborn sleep products despite initial safety guidance, highlighting consumer safety risks from AI chatbots.
- SK Telecom and Upstage launched South Korea's Sovereign AI Cybersecurity Consortium with six security firms and three universities, using A.X K series and Solar models.
- Tom Campbell, CEO of E-ACT, was named Chair of Oak National Academy to oversee digital curriculum and AI tool integration including Aila.
- AI-powered drones from DEEMI, University of Maine, and UMass Amherst aim to cut search and rescue mission time from 72 hours to 7 minutes.
- Google DeepMind's AI model outperformed traditional American and European models during the 2025 hurricane season while tracking Hurricane Isaias.
- Experts propose using AI to support human judgment and building transparent, explainable AI systems to address growing concerns about trust and truth.
Ethereum Researcher Warns AI Could Threaten Wallet Security Within Months
Justin Drake from the Ethereum Foundation warned that AI could break ECDSA, the signature system used by Bitcoin and Ethereum wallets, within months in a worst-case scenario. He cited recent AI advances in mathematical research as a concern. Drake suggested a gradual migration of assets to new addresses that have never signed a transaction. He said Ethereum should speed up its shift to hash-based cryptography. He stressed careful preparation rather than rushing to move funds.
Vitalik Buterin Warns AI Could Weaken Crypto Security in Two Years
Vitalik Buterin warned that AI breakthroughs could weaken lattice-based cryptography and ECDSA within two years. He advised people not to panic but to take the risk seriously. He suggested keeping funds in addresses that have never sent a transaction. He also recommended using offchain confirmations for multisig wallets and sending encrypted notes offchain. For long-term security, he suggested multiplying key sizes by 10.
This Photo Is an AI Fake. Can You Tell?
This article shows an AI-generated image of a FedEx Office made by Google Gemini. The image looks realistic but contains subtle AI mistakes. AI still struggles with perspective, garbled small text, object relationships, and the meaning of words. Experts point out clues like a lanyard not connected to a badge and a nonsensical slogan. The article challenges readers to spot these errors and explains how AI image generation is improving but still has telltale signs.
SK Telecom and Upstage Launch South Korean AI Security Consortium
SK Telecom and Upstage launched the Sovereign AI Cybersecurity Consortium with six security firms and three universities. The group will use SK Telecom's A.X K series and Upstage's Solar models for cybersecurity tasks. The models will be trained on real-world security data and tested through zero-day vulnerability analysis and real-time monitoring. The consortium will also use mutual red-teaming and third-party evaluation from universities. The goal is to improve AI-based cybersecurity through collaboration.
Tom Campbell Named Oak National Academy Chair as AI Work Expands
Tom Campbell, CEO of E-ACT, was named Chair of Oak National Academy. He will oversee the development of the digital national curriculum and the integration of AI tools including Aila. The appointment marks a new phase for AI work in education. Campbell will work closely with the Oak National Academy team to guide innovation. The organization aims to use technology to improve education through AI tools.
AI changes how parents view their kids' creative college degrees
A parent shares their worry about their children's college majors in music production and creative writing. The rise of AI has changed how they think about those creative degrees. The article explores concerns many parents feel about future job prospects in creative fields.
AI era raises questions about trusting narrators and truth
An updated post examines the idea of reliable narrators in the age of AI. AI systems can now generate text, images, and videos, raising concerns about trust and accuracy. The article suggests using AI to support human judgment rather than replace it. Transparent and explainable AI systems are proposed as one solution.
AI powered drones aim to speed up search and rescue missions
DEEMI is partnering with the University of Maine and the University of Massachusetts Amherst to develop AI powered drones for search and rescue. The drones can flag images matching a missing person's description instead of requiring analysts to review hundreds of photos. The team hopes to cut mission time from 72 hours down to 7 minutes. The AI is still being trained and is expected to be operational by the end of the year.
AI chatbots and fake sellers create consumer safety risks online
An article explores how AI chatbots and fake online sellers threaten consumer product safety. A test showed ChatGPT eventually gave unsafe recommendations for newborn sleep products despite initial safety guidance. AI generated social media videos featuring fake people are also being used to sell products. These problems expose weaknesses in traditional product safety enforcement since it is hard to identify anonymous AI generated sellers.
How AI models performed while tracking Hurricane Isaias
The National Hurricane Center leaned on AI models including Google AI to help track Hurricane Isaias. Google AI mainly favored the eastern side of the forecast while traditional physics based models leaned west. The Google Deepmind model outperformed traditional American and European models during the 2025 hurricane season. The National Hurricane Center expects AI models to become more reliable and influential in hurricane forecasting in coming years.
Trump and others blur the line between AI and superintelligence
US President Donald Trump has tried to rename artificial intelligence as super intelligence to make it sound more impressive. He is not the only one shifting the meaning of AI terms over time. Experts like philosopher Nick Bostrom say superintelligence means an intellect that beats all humans in every cognitive field. Current AI systems are still narrow or general AI, not true superintelligence. The article explains how terms like weak AI, strong AI, and AGI have changed as the technology developed.
Sources
- Ethereum Researcher Warns AI Could Threaten Wallet Signature Security Within Months
- AI Could Damage Crypto in 2 Years, Warns Ethereum's Vitalik Buterin
- This Photo Is an A.I. Fake. Can You Tell?
- SK Telecom and Upstage Launch South Korean Security AI Consortium with AhnLab and 10 Other Institutions
- Tom Campbell named Oak National Academy Chair as AI work expands | ETIH EdTech News
- Why AI changed how I think about my kids' arts degrees
- The AI Panic and the Hunt for Reliable Narrators
- AI powered drones are being developed for search and rescue operations
- Article - From AI Chatbots to Fake Sellers: The Consumer Product Safety Challenges of Today's Online Marketplace
- Hurricane Isaias: How AI models have performed while tracking the storm
- AI or ‘super intelligence’? Trump isn’t the only one who is confused
Comments
Please log in to post a comment.