The company said criminals are using Claude not only for technical advice, but also to emotionally pressure victims, in a method it refers to as “vibe hacking”.

In an August 28 report titled Threat Intelligence, Anthropic’s security team, including researchers Ken Lebedev, Alex Moix, and Jacob Klein, explained that “vibe hacking” involves using AI tools to manipulate people’s emotions, gain their trust, and influence their decisions.

Did you know?

Want to get smarter & wealthier with crypto?

Subscribe – We publish new crypto explainer videos every week!

Layer 2 Scaling Solutions Explained With Animations

For example, one hacker reportedly used Claude to help steal private information from 17 different targets, including hospitals, public safety agencies, government offices, and religious groups. The hacker then asked victims for payments in Bitcoin
BTC


$107,488.40

Claude was used to review stolen financial documents, suggest the amount of ransom to be demanded from each victim, and write personalized messages aimed at creating stress or urgency.

Although the attacker’s access to Claude was eventually revoked, the company noted that the situation showed how much easier it has become for people with limited knowledge to create effective malware and avoid detection.

The report also mentioned a separate case involving North Korean IT workers. Anthropic stated that these individuals used Claude to create false identities and pass job interviews for roles at major US tech firms, including some on the Fortune 500 list.

On August 13, ZachXBT revealed how a North Korean hacking group used fake identities and freelance job platforms to secure crypto-related roles. What did the blockchain investigator say? Read the full story.

Anthropic says it has revoked OpenAI’s access to the Claude API after ChatGPT’s engineers were found using Claude’s coding tools.

Claude Code is better than any other coding tool in the AI coding industry, also known as “Vibe coding.”

With Claude, you can create web apps from scratch, and it’s also pretty efficient with infra-related work.

Not just vibe coders who don’t know how to code use Claude, but also professional engineers.

In fact, Claude Code is also used in Claude’s development at Anthropic.

Anthropic offers API access to Claude, which means OpenAI could also access it.

But on Friday, Anthropic confirmed that it revoked OpenAI’s access to Claude Code after the company was found using its coding tools ahead of the GPT-5 launch.

“Claude Code has become the go-to choice for coders everywhere, and so it was no surprise to learn OpenAI’s own technical staff were also using our coding tools ahead of the launch of GPT-5,” Anthropic spokesperson Christopher Nulty noted in a statement to Wired.

It’s unclear how OpenAI engineers were using Claude Code, but it’s possible it was used to improve GPT-5’s coding capabilities.

GPT-5 is expected to launch next week with “auto” and reasoning modes.

Anthropic is secretly working on new models called Claude Sonnet 4 and Opus 4, which are believed to be the company’s most advanced AI models.

This is according to Anthropic web configuration files, which were recently updated with references to Claude Sonnet 4 and Opus 4.

As you can see in the above image showing the internal configuration file, Anthropic has specifically mentioned “Claude 4,” “Claude Sonnet 4,” and “Claude Opus 4.”

These appear to be part of the Claude 4 series, which we think could be Anthropic’s smartest models yet.

The references suggest Claude Sonnet 4 and Opus 4 are being tested or promoted internally, with some warnings like “Not intended for production use” and “strict rate limits,” meaning they’re still under development or early access.

There are also references to features like “show_raw_thinking,” which point to reasoning capabilities.

On May 22, Anthropic is hosting a ‘Code with Claude’ event, but we don’t know if the event is related to the new models.

The reference was included in a court filing by Olivia Chen, a data scientist at the company, as part of its defense against a lawsuit from major music publishers.

Universal Music Group, Concord, ABKCO, and others filed the suit in October 2023, claiming that Anthropic used song lyrics without permission to train its AI model, Claude. The list of artists whose work was allegedly used includes Beyoncé, The Rolling Stones, and The Beach Boys.

Did you know?

Want to get smarter & wealthier with crypto?

Subscribe – We publish new crypto explainer videos every week!

What is NEO in Crypto? Chinese Ethereum Explained (ANIMATED)

Chen’s filing referred to a paper said to be published in The American Statistician. She said it helped show that Claude only repeats copyrighted lyrics in rare situations.

However, at a hearing in San Jose on May 13, the publishers’ lawyer, Matt Oppenheim, said the article was completely made up. He noted that he did not think Chen invented the source herself, but may have used Claude to generate it.

Anthropic’s legal representative, Sy Damle, responded that the issue seemed to be a mistake in referencing, not a made-up source. He also criticized the plaintiffs for bringing it up late in the process.

Judge Susan van Keulen pointed out that a simple mistake is very different from an AI-generated false source. The judge did not allow immediate questioning of Chen but ordered Anthropic to file a formal response by May 15.

On May 12, Jamie Lee Curtis, an American actress, criticized Meta and its CEO, Mark Zuckerberg. What happened? Read the full story.