Back to today’s briefing

Story perspectives

AI Breakthroughs Raise Concerns Over Inaccurate Information

6/3/2025

45 9

Summary

Story summary

Recent breakthroughs in large language models, such as OpenAI's GPT-3 and DeepSeek-R1, have sparked concerns over increased hallucinations—instances where these models produce inaccurate information. Flawed reinforcement learning and memory compression are key culprits. While the EU's AI Act seeks to address these challenges, ensuring accountability remains a significant hurdle in the quest for reliable AI.

1 of 1

Story summary

Recent breakthroughs in large language models, such as OpenAI's GPT-3 and DeepSeek-R1, have sparked concerns over increased hallucinations—instances where these models produce inaccurate information. Flawed reinforcement learning and memory compression are key culprits. While the EU's AI Act seeks to address these challenges, ensuring accountability remains a significant hurdle in the quest for reliable AI.