gpt-3

We tried ChatGPT for vulnerability fixes. Most flaws are too complex for generative AI alone

Security Technology

We tried ChatGPT for vulnerability fixes. Most flaws are too complex for generative AI alone

An experiment with ChatGPT 3.5 found that 80% of code fixes were unusable or introduced new vulnerabilities

clock 12 July 2023 • 4 min read
OpenAI unveils GPT-4

Big Data and Analytics

OpenAI unveils GPT-4

Can handle image inputs and more lengthy requests

clock 15 March 2023 • 3 min read
AI assistants more likely to produce buggy, insecure code

Developer

AI assistants more likely to produce buggy, insecure code

But devs think they've written secure outputs

clock 21 December 2022 • 2 min read
Facebook's new language model has 'high propensity to generate toxic language and reinforce harmful stereotypes'

Applications

Facebook's new language model has 'high propensity to generate toxic language and reinforce harmful stereotypes'

The researchers believe the technology is not yet mature enough for commercial deployment

clock 10 May 2022 • 2 min read