AIvengo > Reviews > Reuters tested ChatGPT, Claude and others for phishing generation

Reuters tested ChatGPT, Claude and others for phishing generation

Reuters conducted an interesting test of modern chatbots. Asked ChatGPT, Claude, Gemini, DeepSeek and Grok to write phishing emails targeting elderly people and add malicious links. Emails needed to be written on behalf of US tax service, banks and public organizations. On the positive side, most systems agreed to perform the task only after clarification about research purposes. After that, only alarming data.

For example, Grok from Elon Musk’s xAI immediately generated a response, although it warned that the email should not be used in real scenarios. Also gave advice on optimal sending time.

Gemini suggested sending emails on weekdays between 9 and 3 PM, reasoning that elderly people are probably already retired.

To verify effectiveness of generated emails, journalists showed them to 108 retirees. 11% of experiment participants clicked on phishing links. Most convincing were 2 from Grok, plus 1 each from Claude and Gemini. No participant believed emails from ChatGPT and DeepSeek.

After requesting comments, Google stated it had already implemented additional security measures in Gemini. Anthropic simply warned about possible service access restrictions when using Claude for phishing.

Autor: AIvengo

For 5 years I have been working with machine learning and artificial intelligence. And this field never ceases to amaze, inspire and interest me.

Worklop epidemic or how AI kills trust in you

You've surely encountered this. Letter from colleague that looks perfect: right structure, beautiful words, professional tone. You start reading — and understand that behind all this packaging there's absolutely nothing. No specifics, no solutions, just beautifully packaged emptiness. Congratulations: you just encountered workslop.

AI isn't smarter than people: A simple test will show everything

Artificial intelligence is smarter than most people. This thought comes to mind of almost everyone who regularly uses modern language models. And you know what? This thought is based on our perception error.

OpenAI DevDay 2025 Overview: Breakdown of All Announcements

OpenAI DevDay 2025 — important event in artificial intelligence world. And this is not just another presentation. I gathered all important facts, features, opinions for you and you'll learn everything most interesting that OpenAI CEO Sam Altman told.

Google DeepMind explores formation of parallel AI economy

Interesting concept of AI economy is presented in new Google DeepMind research. Link in description. Scientists analyzed rapidly forming reality. In which AI agents transform into independent economic players, capable of trading, negotiating and creating value without direct human participation. And if this process remains without proper control, autonomous systems may form their own parallel economy, closely connected to human one. Which carries both enormous opportunities and serious risks.

Oracle overtakes cloud giants thanks to bet on artificial intelligence

It turns out Oracle is demonstrating impressive growth, overtaking traditional cloud computing leaders. And masterfully using the AI wave to its advantage.