We’re in a new era of AI-driven scams

Q: AI-powered cybercrime is accelerating as criminals use LLMs for phishing, deepfakes, and automated v

AI-powered cybercrime is accelerating as criminals use LLMs for phishing, deepfakes, and automated vulnerability scans, making attacks faster and cheaper than ever.

Q: DeepSeek-V4 preview launched claiming to be the most powerful open-source AI platform, rivaling clos

DeepSeek-V4 preview launched claiming to be the most powerful open-source AI platform, rivaling closed-source models from OpenAI and DeepMind, with adaptations for Huawei chips.

Q: Meta is cutting 10% of its workforce (~8,000 jobs) on May 20 to offset AI spending, while OpenAI bro

Meta is cutting 10% of its workforce (~8,000 jobs) on May 20 to offset AI spending, while OpenAI broadly released GPT-5.5 to all ChatGPT users despite cybersecurity concerns.

The Download from MIT Technology Review·Friday, April 24, 2026·6 min read

AI/ML Technology Health

Share𝕏 in

AI Summary

This MIT Technology Review newsletter covers the rise of AI-powered cybercrime, including phishing, deepfakes, and automated vulnerability scans, which are making attacks faster and cheaper. It also examines the growing use of AI in healthcare for notetaking, patient record analysis, and medical imaging, while questioning whether these tools actually improve patient outcomes. Additional news items cover DeepSeek's new V4 model, OpenAI's GPT-5.5 release, Meta's 10% layoffs, and Norway's social media age restrictions.

Key Facts

✓AI-powered cybercrime is accelerating as criminals use LLMs for phishing, deepfakes, and automated vulnerability scans, making attacks faster and cheaper than ever.

✓DeepSeek-V4 preview launched claiming to be the most powerful open-source AI platform, rivaling closed-source models from OpenAI and DeepMind, with adaptations for Huawei chips.

✓Meta is cutting 10% of its workforce (~8,000 jobs) on May 20 to offset AI spending, while OpenAI broadly released GPT-5.5 to all ChatGPT users despite cybersecurity concerns.

Author Takes

SkepticalThe Download from MIT Technology Review

Healthcare AI outcomes

Despite growing evidence that AI tools deliver accurate results in healthcare, there is no good answer yet on whether they actually translate into better health outcomes for patients.

BearishThe Download from MIT Technology Review

Era of free advanced AI

The era of free access to advanced AI is coming to an end as labs face mounting pressure to turn profits.

More from The Download from MIT Technology Review

A new US phone network for Christians aims to block porn and gender-related content

MIT Technology Review's The Download covers a new Christian-marketed US phone network launching with mandatory porn blocks and optional gender-content

May 1

Elon Musk and Sam Altman are going to court over OpenAI’s future

Elon Musk and Sam Altman head to trial this week over OpenAI's future, with Musk seeking $134 billion in damages and a reversion to non-profit status

Apr 28

Want to understand the current state of AI? Check out these charts

Apr 14