Meta blames hallucinations after its AI said Trump rally shooting didn't happen

Meta's AI assistant mistakenly claimed that an attempted assassination of former President Donald Trump did not occur, prompting an apology from the company. Joel Kaplan, Meta's global head of policy, described the AI's response as "unfortunate" and attributed the error to the technology's tendency to produce "hallucinations," a common issue in generative AI systems. Initially, Meta had restricted its AI from discussing the assassination attempt but lifted that ban after public scrutiny. Despite this, the AI continued to generate incorrect responses, including the assertion that the event did not happen. Kaplan acknowledged the ongoing challenge of ensuring AI systems provide accurate information, especially regarding real-time events. He emphasized that all generative AI models can produce inaccurate outputs and that Meta is committed to addressing these issues as the technology evolves. The incident also coincided with Google facing accusations of censoring search results related to the assassination attempt, which Trump criticized on social media. The tech industry continues to grapple with the challenges posed by generative AI, particularly in limiting the spread of misinformation.