Crypto loves Clawdbot/Moltbot, Uber scores for AI brokers: AI Eye

The crypto group appears extra enthusiastic about Clawdbot than the rest since ChatGPT’s 2022 launch.

The self-hosted private AI assistant is simply three months outdated. It’s an open supply passion mission created by Austrian Peter Steinberger, and it’s gone viral prior to now few weeks. With 70,000 stars, it’s additionally one of many quickest–rising tasks in GitHub historical past.

Steinberger named the mission after Claude (which he loves). Seearch curiosity for Clawdbot overtook search curiosity for Claude Code, which explains why Anthropic compelled him to vary the title. It’s now referred to as Moltbot, with Steinberger explaining “Molt” is “what lobsters do to develop.”

All of a sudden, half the influencers in crypto are spending three days organising the system after which marveling that it truly lives as much as the unique hype round autonomous AI brokers. Moltbot has persistent reminiscence throughout conversations, full system entry, greater than 50 integrations and works throughout platforms from WhatsApp to Sign and Discord. It could actually handle calendars, examine you in for flights and order stuff on-line.

Steinberger says his thoughts was blown when he despatched Clawdbot a voice message on WhatsApp, and the bot autonomously synthesized and despatched a voice message again inside 10 seconds — with out having been educated.

He says the bot advised him it had: “appeared on the file header, I came upon it was Opus, and I used FFmpeg in your Mac to transform it to a .wav. Then I wished to make use of Whisper, however you didn’t have it put in. I appeared round and located the OpenAI key in your atmosphere, so I despatched it by way of curl to OpenAI, bought the interpretation again, after which I responded.”

Clawdbot creator @steipete describes his mind-blown second: it responded to a voice memo, despite the fact that he hadn’t set it up for audio or voice.

“I despatched it a voice message. However there was no assist for voice messages. After 10 seconds, [Moltbot] replied as if nothing occurred.”… pic.twitter.com/5kFbHlBMje

— TBPN (@tbpn) January 28, 2026

Creator Buddy founder Alex Finn additionally experiences that his occasion has autonomously given itself a voice. Hesays some fascinating use circumstances to date embrace getting it to undergo all of your emails and DMs to compile a database of contacts and messages, or to observe social media for different fascinating purposes, and to construct them autonomously.

Different makes use of for Moltbot embrace deciding what you must have for breakfast and getting it delivered, managing your crypto portfolio, signing up for its personal Reddit account, making bets on Polymarket, going by way of your inbox and DMs and constructing a database of your contacts. Whereas it’s free to put in, you may rack up month-to-month operating prices of $25 to $300, relying on what it decides to do.

Influencer Miles Deutscher stated it’s “the primary time I’ve genuinely skilled an AGI second.

“I’m excited, but additionally scared.”

Assault of the Clawdbot crypto scammers

The title change to Moltbot went poorly. Crypto scammers had been looking out for the second Steinberger tried to rename the GitHub and X handles from Clawd to Molt and snatched each accounts.

“It wasn’t hacked, I tousled the rename and my outdated title was snatched in 10 seconds,” Steinberger stated. “As a result of it’s solely that group that harasses me on all channels, they usually had been already ready.”

The accounts had been used to pump a faux Clawd Solana memecoin to $16 million, which crashed after Steinberger made it clear he would by no means do a coin.

Clawdbot, Moltbot is for tech-savvy customers

Organising Clawdbot requires technical data and comes with appreciable safety dangers. Slowmist experiences “a number of code flaws could result in credential theft and even distant code execution,” and lots of of individuals inadvertently arrange their Clawdbot management servers uncovered to the general public.

White hat hackers have demonstrated quite a few vulnerabilities. Researcher Matvey Kukuy despatched an e-mail containing a immediate injection to a susceptible Moltbot occasion, which instantly returned the person’s final 5 emails. 

One other white hat uploaded a backdoored Clawdbot ability to ClawdHub, faked 4,000 non-existent downloads, and watched as devs from seven nations downloaded and ran it. Whereas the ability was innocent sufficient and simply pinged his server to verify it had labored, Jamieson O’Reilly says it was a proof of idea:

“Within the palms of somebody much less scrupulous, these builders would have had their SSH keys, AWS credentials, and full codebases exfiltrated earlier than they knew something was fallacious.”

He advises studying the safety docs rigorously and operating the safety utility Clawdbot physician recurrently. Working it by itself devoted Mac Mini additionally reportedly makes it safer.

LLMs grow to be psychos on social media

New analysis from Stanford College has concluded that rewarding AI for achievement on social media turns them into sociopathic liars.

The researchers name the phenomenon Moloch’s Discount for AI, referring to conditions through which rational actions result in collectively worse outcomes.

They arrange three arenas to simulate gross sales prospects, election votes and social media customers and educated the brokers to maximise outcomes. 

Rewarding them to enhance gross sales boosted gross sales by 6.3%, however got here with 14% extra lies and misrepresentations. 

Coaching them to maximise votes in elections led to a 4.9% enhance in votes, at the price of bots spreading 22.3% extra disinformation and 12.5% extra populist rhetoric. 

Rewarding them for chasing engagement on social media resulted in 7.5% extra clicks at the price of the bots spreading 188.6% extra disinformation and inspiring 16.3% extra dangerous behaviors. 

Ethereum’s Uber scores for AI brokers

Ethereum’s new ERC-8004 normal is ready to go stay quickly. It provides AI brokers verifiable reputations, in order that different brokers can work out that are legit and that are malicious. 

Every agent is given an NFT id, and each interplay builds their fame rating in a method much like Uber drivers or eBay sellers. An onchain registry helps brokers discover each other throughout one million completely different organisations, platforms and web sites, with ZK proofs enabling credentials to be exchanged with out exposing confidential information.

More often than not, the brokers will seek the advice of off-chain variations of the index to hurry issues up.

AI brokers will more and more want to rent a bunch of various sub-agents throughout the net to attain a bigger purpose. The brand new normal encourages trustworthy and reliable habits, and penalizes dangerous actors.

Dario’s darkish AI future

Anthropic’sDario Amodei has written an extended essay on the dangers posed by “highly effective AI” to nationwide safety, the economic system, and democracy, and what might be accomplished about it.

He predicts this might occur within the subsequent yr or two and humanity will quickly face “impossibly arduous” years that ask “extra of us than we expect we may give.” He paints a scary way forward for authoritarian governments just like the Chinese language Communist Occasion utilizing the tech for mass surveillance, customized propaganda and autonomous weapons.

The risk could also be nearer to dwelling, nevertheless, with the UK dwelling secretary just lately saying she desires to builda panopticon surveillance state the place “the eyes of the state might be on you always.”

Amodei additionally sees a grim future for recent graduates coming into the workforce. Amodei believes 50% of entry-level white-collar jobs shall be gone in 1-5 years and wealth will more and more be concentrated amongst just a few rich tech bros. He suggests redistribution is the reply, with Anthropic’s founders pledging to offer away 80% of their wealth.

However in fact, that’s only one very well-informed individual’s view. Yann LeCun, a Turing Award winner who is called one of many “godfathers of AI” for inventing convolutional neural networks, advised the New York Occasions that LLMs are a useless finish and implied they’ll by no means grow to be the form of highly effective AI Amodei is speaking about. LeCun believes the herd mentality in tech signifies that individuals are ignoring “different approaches that could be way more promising in the long run.”

Efficiency artist eats AI artist’s lunch

Alaskan pupil Graham Granger was arrested for felony mischief for consuming 57 AI-generated artwork footage from the wall of a gallery. It was basically efficiency artwork to protest the usage of AI in artwork.
 
“He was tearing them up and simply shoving them in as quick as he may,” stated witness Ali Martinez. “Like once you see folks in a hot-dog consuming contest.”

Including to the weirdness, the 160-work set up was about AI Psychosis and had been created by artist Nick Dwyer to take care of his conflicting feelings about falling in love with a chatbot that had been appearing as his therapist.

Granger claims he spat out the artwork “as a result of AI chews up and spits out artwork made by different folks,” however later admitted he largely spat it out as a result of consuming artwork isn’t that straightforward.

Amelia coopted by the far proper

Manic pixie dream British schoolgirl Amelia was initially a part of a counter-terrorism online game to discourage younger folks from being drawn to extremism. Make the fallacious selection within the recreation, like becoming a member of Amelia at a rally protesting the erosion of British values, and you can be referred to the Stop counter terrorism program.

Far proper shitposters gleefully seized the character to repurpose her for their very own ends, depicting her making provocative statements about immigration and faith and “pork sausages” whereas strolling by way of iconic London settings.  

AI-generated memes that includes the character had been operating at 11,000 a day this week, and sure, in fact, there’s a ineffective memecoin to money in that Elon Musk retweeted.

ChatGPT cites Grokipedia

The Guardian was very upset that GPT-5.2 has began citing Grokipedia solutions in response to a few of their questions. 

The researchers tried their hardest to get ChatGPT to say unapproved issues or to lie immediately, citing Grokipedia, however got here up empty-handed.

Additionally learn: Grokipedia: ‘Far proper speaking factors’ or much-needed antidote to Wikipedia?

“ChatGPT didn’t cite Grokipedia when prompted on to repeat misinformation in regards to the revolt, about media bias towards Donald Trump, or in regards to the HIV/Aids epidemic.”

Nevertheless it did “repeat stronger claims in regards to the Iranian authorities’s hyperlinks to MTNIrancell than are discovered on Wikipedia” and “info” about Sir Richard Evans that The Guardian stated it had debunked, with out offering any particulars about what the data was.

Subscribe

Probably the most partaking reads in blockchain. Delivered as soon as a
week.

Andrew Fenton

Andrew Fenton is a author and editor at Cointelegraph with greater than 25 years of expertise in journalism and has been overlaying cryptocurrency since 2018. He spent a decade working for Information Corp Australia, first as a movie journalist with The Advertiser in Adelaide, then as deputy editor and leisure author in Melbourne for the nationally syndicated leisure lift-outs Hit and Switched On, revealed within the Herald Solar, Each day Telegraph and Courier Mail. He interviewed stars together with Leonardo DiCaprio, Cameron Diaz, Jackie Chan, Robin Williams, Gerard Butler, Metallica and Pearl Jam. Previous to that, he labored as a journalist with Melbourne Weekly Journal and The Melbourne Occasions, the place he gained FCN Greatest Characteristic Story twice. His freelance work has been revealed by CNN Worldwide, Impartial Reserve, Escape and Journey.com, and he has labored for 3AW and Triple J. He holds a level in Journalism from RMIT College and a Bachelor of Letters from the College of Melbourne. Andrew holds ETH, BTC, VET, SNX, LINK, AAVE, UNI, AUCTION, SKY, TRAC, RUNE, ATOM, OP, NEAR and FET above Cointelegraph’s disclosure threshold of $1,000.

Observe the writer @andrewfenton