Ethereum Foundation researcher Justin Drake warned that artificial intelligence could break the ECDSA cryptography securing crypto wallets within months, rather than years. He urged users to gradually move assets to fresh addresses that hide public keys behind hashes, a strategy he called "bunker mode." His warning followed OpenAI's release of hundreds of new mathematical findings in October. Drake explicitly cautioned against rushing the migration, saying a hasty move would do more harm than good.
Not everyone agreed with the urgency. Coinbase cryptography head Yehuda Lindell said there is no evidence that security assumptions have weakened. Ethereum founder Vitalik Buterin echoed a measured tone, urging people to take the risk seriously but not panic yet. The debate highlights uncertainty about when — or if — AI will pose a real threat to digital-asset security.
In a separate development, Anthropic agreed to a $1.5 billion settlement in the Bartz v. Anthropic lawsuit over using copyrighted books to train its AI models. Among the works included was Tufts Associate Professor Craig Cipolla's 2013 book "Becoming Brothertown." Cipolla said the infringement was widespread across academia. The settlement avoided a verdict but left legal questions about future AI development unresolved.
On the education front, Rochester Institute of Technology introduced a course called AI and Copyright for Creatives. Adjunct Professor Rachel DiNunzio teaches art students about copyright law and how AI affects creative work. Students learn both defensive and offensive tactics to protect their art. The course will continue in spring 2027 and is open to graduate and undergraduate students in the College of Art and Design.
Oracle is putting AI tools to work at scale, with over 130,000 employees using ChatGPT and 95,000 using Codex. Recruiters now build market intelligence tools in minutes instead of days, and engineers report faster, more consistent results. Management stresses that humans still oversee all AI-generated work, keeping a hand on the lever even as automation speeds things up.
Copper demand is rising sharply as AI data centers expand. Coppernico Metals founder Ivan Bebek said tech giants including Meta, Microsoft, Alphabet, and Amazon are securing copper supplies to support the infrastructure. Concerns are growing that copper could hit $10 per pound due to the wiring needs of new facilities.
AI is also making its way into political advertising. A Wesleyan Media Project report found nearly 170 AI-generated political ads through early September, backed by over $80 million in spending. Researchers identified satirical depictions and hyperrealistic deepfakes of candidates from both parties. Almost 70 percent of the ads lacked disclaimers about AI use because no federal law requires disclosure.
Key Takeaways
- Ethereum researcher Justin Drake warns AI could break crypto wallet ECDSA security within months, not years
- Drake recommends a gradual migration to addresses with hidden public keys, cautioning against rushed transfers
- Coinbase cryptography head Yehuda Lindell says there is no evidence security assumptions have weakened
- Anthropic agreed to a $1.5 billion settlement in a copyright lawsuit over using copyrighted books to train AI models
- RIT launched a new AI and Copyright for Creatives course for art students, running through spring 2027
- Oracle reports over 130,000 employees use ChatGPT and 95,000 use Codex to speed up daily tasks
- Meta, Microsoft, Alphabet, and Amazon are securing copper supplies as AI data center demand drives prices toward $10 per pound
- Nearly 170 AI-generated political ads appeared ahead of midterm elections, with over $80 million in spending
- About 70 percent of AI campaign ads lacked disclaimers because no federal law requires disclosure
- AgentMemGate improved conversational AI task accuracy from 65% to 95% by preventing unconfirmed plans from being stored as facts
Justin Drake urges crypto bunker mode as AI may break wallet security soon
Ethereum researcher Justin Drake warned that AI could break the ECDSA cryptography securing crypto wallets within months in a worst-case scenario. He recommended users gradually move assets to fresh addresses that hide public keys behind hashes. His warning followed OpenAI's release of hundreds of new mathematical findings. Drake stressed that a rushed migration would do more harm than good.
Ethereum researcher warns AI threat to wallet security could emerge within months
Ethereum Foundation researcher Justin Drake urged the crypto industry to prepare for a potential AI breakthrough that could undermine wallet security. He said ECDSA could break before qday, possibly in months rather than years, allowing fast private key recovery. He recommended moving funds to addresses whose public keys remain hidden behind a hash. Large holders should lead the migration carefully without panicking.
Security researcher warns AI could crack crypto wallets in months
Justin Drake issued a viral warning urging Bitcoin and crypto holders to prepare for an AI security threat that could emerge within months. He called for a controlled migration to addresses with hidden public keys and said signing a transaction exposes keys that should be moved afterward. He noted the warning followed OpenAI's October 6 mathematics release. Drake explicitly cautioned against rushed transfers.
AI could breach crypto wallet security before quantum computers, researcher says
Justin Drake of the Ethereum Foundation said advances in AI could undermine ECDSA security in digital-asset wallets within months rather than years. He said the threat could arrive before quantum computers become powerful enough to break the same cryptography. The warning was cited by Wu Blockchain on October 8. He urged the industry to plan for what he called bunker mode.
Tufts professor's book among pirated works in Anthropic copyright settlement
Anthropic agreed to a $1.5 billion settlement in the Bartz v. Anthropic lawsuit over using copyrighted books to train AI models. Associate Professor Craig Cipolla of Tufts University said his 2013 book Becoming Brothertown was included in Anthropic's dataset without permission. Cipolla said the infringement was widespread across academia. The settlement avoided a verdict, leaving legal questions about future AI development unresolved.
RIT launches AI and copyright course for creative students
RIT introduced a new course called AI and Copyright for Creatives to help art students understand their legal rights. Adjunct Professor Rachel DiNunzio teaches students about copyright law and how AI affects creative work. Students learn defensive and offensive tactics to protect their art. The course will continue in spring 2027 and is open to graduate and undergraduate students in the College of Art and Design.
Oracle uses ChatGPT and Codex to speed up work
Oracle uses ChatGPT Work and Codex to help employees complete tasks that once took days. Over 130,000 employees use ChatGPT and 95,000 use Codex across the company. Recruiters now build market intelligence tools in minutes instead of days. Engineers and business users also benefit from faster, more consistent results. Management stresses that humans still oversee all AI-generated work.
Economics framework proposed for evaluating language models
Researchers suggest using stated-preference economics to evaluate large language models on questions with no clear answer. The framework uses concepts like validity, reliability, and incentive compatibility from economics. The team tested six models using a water-quality valuation survey. Two older models failed basic tests while newer models passed most tests. Passing tests shows answers are coherent but not necessarily correct.
AgentMemGate prevents AI memory errors from unconfirmed plans
AgentMemGate is a new tool that stops conversational AI assistants from treating plans as facts too early. When users mention possible future actions, the system classifies them as speculation or completed events. Speculations are kept out of memory until confirmed. Testing showed the tool eliminated contamination in most cases and improved task accuracy from 65% to 95%. The research team also released datasets and evaluation code.
Amdocs lands 10-year AI telecom deal with VEON
Amdocs signed a 10-year agreement with VEON to modernize technology operations using its aOS platform and QazCode expertise. The deal strengthens Amdocs' role in AI-driven telecom transformation across Eurasia. Combined with a similar Liberty Latin America contract, it shows a trend toward long-term AI platform projects. Amdocs forecasts revenue of US$5.2 billion and earnings of US$945.9 million by 2029.
AI could threaten crypto wallet security warn experts
Ethereum researcher Justin Drake warned that AI breakthroughs could weaken crypto wallet cryptography. He said the risk could emerge within months. Coinbase cryptography head Yehuda Lindell pushed back and said there is no evidence that security assumptions have weakened. Ethereum founder Vitalik Buterin urged taking the risk seriously but not panicking yet.
AI data center growth may cause copper shortages
Coppernico Metals founder and CEO Ivan Bebek discussed rising copper demand driven by AI data center construction. Tech giants including Meta, Microsoft, Alphabet and Amazon are securing copper supplies. Concerns grow that copper could hit $10 per pound due to AI wiring needs.
LGPS investors navigate the growing AI investment debate
The Room 151 LGPS Equities Roundtable in York explored how institutional investors are handling AI challenges. Michael Feldman from Clearbridge Investments shared a positive outlook on AI. He noted that AI remains large and continues to grow. The discussion covered many facets of the debate around AI in investment markets.
AI campaign ads on the rise ahead of midterm elections
A Wesleyan Media Project report found nearly 170 AI generated political ads this year through early September. Over $80 million in spending backed these ads. Researchers identified satirical depictions and hyperrealistic deepfakes of candidates from both parties. Almost 70 percent of the ads lacked disclaimers about AI use because no federal law requires disclosure.
Sources
- Justin Drake urges crypto ‘bunker mode,’ as AI could break wallet security within months
- AI | Leading Ethereum Foundation Researcher Says Potential AI Threat to Wallet Security Could Emerge ‘Within Months’
- Move Your Bitcoin? Security Researcher's Viral Warning Says AI Could Crack Crypto in Months
- AI Could Breach Crypto Wallet Security Before Quantum Computers, Ethereum Researcher Says
- Tufts professor’s book among illegally pirated works in Anthropic copyright settlement
- New RIT course on AI and copyright helps creative students feel empowered
- How Oracle turns days of work into minutes with ChatGPT and Codex
- Validity Without Ground Truth: What Stated-Preference Economics Offers the Evaluation of Language Models
- AgentMemGate: Addressing Speculation Contamination in Conversational Assistant Memory
- AI Telecom Deal Might Change The Case For Investing In Amdocs Stock (DOX)
- Could AI break crypto wallet security?
- AI data center buildout could lead to copper shortages
- How are LGPS investors grappling with the AI race?
- A Connecticut report shows uptick in AI campaign ads nationally
Comments
Please log in to post a comment.