The paradox of curiosity in the age of AI
Curiosity drives technology research and development, but does it drive and magnify the risks of AI itself? And what happens if AI develops its own curiosity?…
By Sam Jeans · 11 min read
AI risks
Western, Chinese scientists draw AI development ‘red lines’
Last week, leading AI scientists met at the second International Dialogue on AI Safety in Beijing to agree on ‘red lines’ for AI development to mitigate…
By Eugene van der Watt · 3 min read
US-commissioned report says AI poses “extinction-level threat”
A US government-commissioned report says that new AI safety measures and policies are needed to prevent an “extinction-level threat to the human species.” The report, which…
By Eugene van der Watt · 3 min read
Researchers jailbreak LLMs by using ASCII art in prompts
Researchers developed a jailbreak attack called ArtPrompt, which uses ASCII art to bypass an LLM’s guardrails. If you remember a time before computers could handle graphics,…
By Eugene van der Watt · 2 min read
Microsoft’s Copilot goes into God mode and it’s a little scary
Copilot users shared disturbing interactions that they had with Microsoft’s AI assistant in response to prompts that referred to the chatbot as “SupremacyAGI”. Copilot may simply…
By Eugene van der Watt · 2 min read
IBM Security shows how AI can hijack audio conversations
IBM Security published research on its Security Intelligence blog to show how AI voice clones could be injected into a live conversation without the participants realizing…
By Eugene van der Watt · 2 min read
Does AI display racial and gender bias when evaluating images?
Researchers from the National Research Council Canada performed experiments on four large vision-language models (LVLM) to see if they displayed racial and gender bias. AI models…
By Eugene van der Watt · 4 min read
AI models have a tendency to escalate wargame scenarios, says study
AI chatbots, particularly those developed by OpenAI, tend to select aggressive tactics, including the use of nuclear weaponry, according to a new study. The research conducted…
By Sam Jeans · 4 min read
OpenAI says GPT-4 could help you make a bioweapon, maybe
The RAND think tank published a report in October saying AI “might” help to build a bioweapon, and then published a subsequent report to say it…
By Eugene van der Watt · 2 min read
Deep fake audio getting easier to make, harder to detect
Fake AI-cloned voices made the news recently because of a “Biden” robocall, but ordinary people are being affected as the technology becomes more accessible and harder…
By Eugene van der Watt · 3 min read