Anthropic researchers predict human-level AI could arrive within two years, with biology becoming the most important area for development. The team also warns that U.S. lab infrastructure is falling behind China, and that faster intelligence growth could lower barriers for harmful uses in biology and cybersecurity.
In a separate development, the Arizona Court of Appeals overturned a 10-year manslaughter sentence for Gabriel Paul Horcasitas after ruling that an AI-generated video of the deceased victim was unreliable. The video, created by Pelkey's family using photos and voice recordings, was deemed to blur the line between family beliefs and the victim's actual voice. This case may be the first in U.S. courts to address this kind of AI evidence.
NEAR co-founder Illia Polosukhin shared 13 application ideas on X, including NFT-based AI pets called AI Tamagochi, universal checkout widgets for merchants, and cross-chain event deposit systems. He also proposed private AI tools for trading, code reviews, and medical second opinions.
The UK's Financial Conduct Authority warns that the rapid pace of AI development is exposing weak spots in firms' cybersecurity. Many companies use AI to automate tasks, but the technology also creates new risks that existing controls cannot keep up with. The regulator urges firms to understand AI risks and take action now rather than waiting for a major incident.
David Martin from Boston Consulting Group warned about over-reliance on AI, identifying two specific habits that can weaken critical thinking skills. He emphasized the need to stay engaged rather than depend too heavily on automated outputs as companies increasingly integrate AI tools into daily work.
Research shows AI both creates and eliminates jobs simultaneously. Since 2021, highly substitutable work has grown 3.3 percentage points faster per year than non-substitutable work. The study classifies U.S. work tasks separately to show both effects, explaining why estimates of AI's employment impact differ so widely.
Two PlusAI trailers were stolen from a warehouse in Fremont, California, attracting attention because of their prominent Nvidia and PlusAI branding. Each trailer contained about 20,000 pounds of sand used for testing autonomous trucks. Police recovered both trailers hours later after a friend spotted them in nearby Newark.
Graphify Labs built a tool using knowledge graphs to help AI agents understand business concepts, with over 123,000 GitHub stars and 7.9 million PyPI downloads. A McKinsey survey found that while 80% of companies saw productivity gains from AI, only 37% saw improved earnings.
Two letters to the editor discuss risks from advanced AI systems. Loren McArthur argues superintelligent AI cannot be stopped by any human institution and calls for a ban on developing such technologies. Another reader compares AI to nuclear weapons, noting that AI lacks physical scarcity and can be hidden in ordinary software.
Key Takeaways
- Anthropic researchers predict human-level AI could arrive within two years, with biology identified as the most important area for AI development.
- U.S. lab infrastructure is falling behind China, according to Anthropic's reinforcement learning leaders.
- An Arizona appeals court overturned a 10-year manslaughter sentence because an AI-generated video of the deceased victim was ruled unreliable.
- NEAR co-founder Illia Polosukhin proposed 13 AI application ideas including NFT-based AI pets and cross-chain payment systems.
- The UK's FCA warns that AI is exposing cybersecurity weak spots that existing controls cannot keep up with.
- Boston Consulting Group warns that over-reliance on AI can weaken critical thinking skills.
- Since 2021, highly substitutable work has grown 3.3 percentage points faster per year than non-substitutable work due to AI.
- Two Nvidia-branded PlusAI trailers containing 20,000 pounds of sand each were stolen in Fremont, California.
- Graphify Labs has over 123,000 GitHub stars; 80% of companies saw productivity gains from AI but only 37% saw improved earnings.
- Readers debate stronger AI regulation, with one comparing AI risks to nuclear weapons and another calling for a development ban.
Arizona road rage killer to be resentenced after AI victim video ruled improper
Gabriel Paul Horcasitas was convicted of manslaughter for killing Christopher Pelkey during a road rage shooting in 2021. He received a 10-year prison sentence. An appeals court ruled the sentence must be thrown out because the court allowed an AI-generated video of Pelkey to be shown. The court said the video was not reliable and unfairly influenced the judge.
Arizona court tosses sentence in case involving AI video of dead victim
Gabriel Paul Horcasitas had his 10-year manslaughter sentence overturned by the Arizona Court of Appeals. The court ruled that an AI video showing the deceased victim speaking was unreliable. The video was created by Pelkey's family using photos and voice recordings. The family wanted to humanize Pelkey for the judge. The case may be the first in U.S. courts to involve this kind of AI evidence.
Road rage killer's sentence quashed over AI victim video in court
Gabriel Paul Horcasitas killed Christopher Pelkey in a 2021 road rage incident at a red light in Chandler, Arizona. He was sentenced to 10.5 years in prison. An appeals court overturned the sentence because the judge allowed an AI video of Pelkey to be played. The court said the video blurred the line between the family's beliefs and the victim's actual voice. The judge who oversaw the case had praised the AI video.
Anthropic researchers predict human-level AI in two years
Anthropic's reinforcement learning leaders say artificial general intelligence could arrive within two years. They believe biology will be the most important area for AI. They also warn that U.S. lab infrastructure is falling behind China. The researchers say scaling up lab space and supply chains is critical. They also note that faster intelligence growth could lower barriers for harmful uses in biology and cybersecurity.
AI agents can write code but may not understand your business
Graphify Labs built a tool that uses knowledge graphs to help AI agents understand business concepts. The company has over 123,000 GitHub stars and 7.9 million PyPI downloads. A McKinsey survey found that while 80% of companies saw productivity gains from AI, only 37% saw improved earnings. Graphify aims to help AI agents understand shared business terms across departments.
NEAR co-founder proposes AI pets and cross-chain payment ideas
NEAR co-founder Illia Polosukhin shared 13 application ideas on X. The proposals include NFT-based AI pets called AI Tamagochi, universal checkout widgets for merchants, and cross-chain event deposit systems. He also suggested private AI tools for trading, code reviews, and medical second opinions. The ideas aim to help developers build products and reach users more effectively.
Thieves steal Nvidia-branded trailers, find 20 tons of sand
Two PlusAI trailers were stolen from a warehouse in Fremont, California. The trailers had prominent Nvidia and PlusAI branding, which likely attracted the thieves. Each trailer contained about 20,000 pounds of sand used for testing autonomous trucks. Police recovered both trailers hours later after a friend spotted them in nearby Newark.
AI creates and eliminates jobs at the same time
Research shows AI both adds and removes jobs simultaneously. Since 2021, highly substitutable work has grown 3.3 percentage points faster per year than non-substitutable work. The study classifies US work tasks separately to show both effects. The findings explain why estimates of AI's employment impact differ so widely.
Readers debate how to control dangerous AI development
Two letters to the editor discuss risks from advanced AI systems. Loren McArthur argues that superintelligent AI cannot be stopped by any human institution and calls for a ban on developing such technologies. Another reader compares AI to nuclear weapons, noting that AI lacks physical scarcity and can be hidden in ordinary software. Both letters urge stronger regulation beyond voluntary company commitments.
BCG exec warns AI habits can weaken critical thinking
David Martin from Boston Consulting Group warned about over-reliance on AI. He identified two specific habits that can weaken a person's critical thinking skills. The advice comes as companies increasingly integrate AI tools into daily work. Martin emphasized the need to stay engaged rather than depend too heavily on automated outputs.
FCA warns AI is exposing cyber risks firms struggle to manage
The UK's Financial Conduct Authority says the fast pace of AI development is revealing weak spots in firms' cybersecurity. Many companies use AI to automate tasks and improve efficiency, but the technology also creates new risks. The FCA warns that existing controls cannot keep up with these evolving threats. The regulator urges firms to understand AI risks and take action now rather than waiting for a major incident.
Sources
- US road rage killer's sentence quashed because AI video of victim was shown in court
- Sentence tossed in Arizona case where deceased victim was depicted speaking in AI-generated video
- Road rage killer has sentence quashed over AI video
- Anthropic RL leads say AGI arrives in two years
- Your AI Agents Can Write Code. Can They Understand Your Business?
- NEAR co-founder pitches AI pets, private trading and cross-chain payments
- Nvidia-branded trailers stolen in California, thieves discover 20 tons of sand instead AI hardware
- Hiding in the mean: AI is adding jobs and removing them at the same time
- Opinion | Controlling Risks Posed by A.I. and Other Threats to Humanity
- BCG exec shares 2 AI habits that can weaken critical thinking
- FCA warns AI may reveal more cyber vulnerabilities firms can cope with
Comments
Please log in to post a comment.