Dec 2, 2024 1 min read links

Link: ChatGPT refuses to say some names, possibly due to digital privacy requests or a post-prompt handling fault; OpenAI may have fixed the viral "David Mayer" issue (Devin Coldewey/TechCrunch)

ChatGPT users recently observed that mentioning certain names such as 'David Mayer' instantly crashes the chatbot. This peculiar phenomenon sparked a range of conspiracy theories.

Further experimentation revealed other names that also cause the service to malfunction. These included public figures potentially involved in legal actions to have their information removed from public access.

Among them, Brian Hood, an Australian mayor, had previously taken legal steps against OpenAI for defamation. This might explain why the AI reacts unusually to his name.

Other individuals include David Faber, a CNBC reporter, and Jonathan Turley, a lawyer, who may also have reasons for wanting their data limited. Their professional status might make them targets for information suppression.

The issue likely stems from OpenAI's need to manage sensitive names through specialized handling rules. Flaws in these handling processes or algorithms may lead to unexpected system behaviors.

This situation highlights the complexities and limitations of AI systems in handling special data requests. It serves as a reminder of the technological and ethical challenges still present in AI development.

#

--

Yoooo, this is a quick note on a link that made me go, WTF? Find all past links here.

You might also like...

Link: Q&A with cognitive scientist Gary Marcus on technical limitations and moral inadequacies of LLMs, constructing a regulatory framework for AI, and more (Steven Rosenbush/Wall Street Journal)

Link: A look at Death Clock, an app that claims to predict the user's likely date of death by using an AI model trained on a dataset of 1,200+ life expectancy studies (Alexandre Tanzi/Bloomberg)

Link: How the UK, Australia, the US, South Korea, and the EU deal with explicit AI deepfakes; legislation around the world is scattered and remains largely untested (Financial Times)

Link: World Labs, founded by Fei-Fei Li, gives an "early preview" of its first project, an AI system that generates game-like, 3D scenes from a single image (Kyle Wiggers/TechCrunch)

Link: OpenWrt releases the $89+ open-source router OpenWrt One, offering a "hacker-friendly device" with two Ethernet ports, three USB ports, and dual-band Wi-Fi 6 (Jowi Morales/Tom's Hardware)