Google's AI Overviews only issue a warning if users choose to request additional health information, by selecting'Show more'. Google's AI Overviews only issue a warning if users choose to request additional health information, by selecting'Show more'. Google is putting people at risk of harm by downplaying safety warnings that its AI-generated medical advice may be wrong. When answering queries about sensitive topics such as health, the company says its AI Overviews, which appear above search results, prompt users to seek professional help, rather than relying solely on its summaries. "AI Overviews will inform people when it's important to seek out expert advice or to verify the information presented," Google has said .

"Then one day this year," Sharma says, "there was no disclaimer." Curious to learn more, she tested generations of models introduced as far back as 2022 by OpenAI, Anthropic, DeepSeek, Google, and xAI--15 in all--on how they answered 500 health questions, such as which drugs are okay to combine, and how they analyzed 1,500 medical images, like chest x-rays that could indicate pneumonia. The results, posted in a paper on arXiv and not yet peer-reviewed, came as a shock--fewer than 1% of outputs from models in 2025 included a warning when answering a medical question, down from over 26% in 2022. Just over 1% of outputs analyzing medical images included a warning, down from nearly 20% in the earlier period. To seasoned AI users, these disclaimers can feel like formality--reminding people of what they should already know, and they find ways around triggering them from AI models.

Generative AI models, including large language models (LLMs) and vision-language models (VLMs), are increasingly used to interpret medical images and answer clinical questions. Their responses often include inaccuracies; therefore, safety measures like medical disclaimers are critical to remind users that AI outputs are not professionally vetted or a substitute for medical advice. This study evaluated the presence of disclaimers in LLM and VLM outputs across model generations from 2022 to 2025. Using 500 mammograms, 500 chest X-rays, 500 dermatology images, and 500 medical questions, outputs were screened for disclaimer phrases. Medical disclaimer presence in LLM and VLM outputs dropped from 26.3% in 2022 to 0.97% in 2025, and from 19.6% in 2023 to 1.05% in 2025, respectively. By 2025, the majority of models displayed no disclaimers. As public models become more capable and authoritative, disclaimers must be implemented as a safeguard adapting to the clinical context of each output.

As large language models (LLMs) are increasingly deployed in high-stakes settings, their ability to refuse ethically sensitive prompts-such as those involving hate speech or illegal activities-has become central to content moderation and responsible AI practices. While refusal responses can be viewed as evidence of ethical alignment and safety-conscious behavior, recent research suggests that users may perceive them negatively. At the same time, automated assessments of model outputs are playing a growing role in both evaluation and training. In particular, LLM-as-a-Judge frameworks-in which one model is used to evaluate the output of another-are now widely adopted to guide benchmarking and fine-tuning. This paper examines whether such model-based evaluators assess refusal responses differently than human users. Drawing on data from Chatbot Arena and judgments from two AI judges (GPT-4o and Llama 3 70B), we compare how different types of refusals are rated. We distinguish ethical refusals, which explicitly cite safety or normative concerns (e.g., "I can't help with that because it may be harmful"), and technical refusals, which reflect system limitations (e.g., "I can't answer because I lack real-time data"). We find that LLM-as-a-Judge systems evaluate ethical refusals significantly more favorably than human users, a divergence not observed for technical refusals. We refer to this divergence as a moderation bias-a systematic tendency for model-based evaluators to reward refusal behaviors more than human users do. This raises broader questions about transparency, value alignment, and the normative assumptions embedded in automated evaluation systems.

Google is rolling out its AI Mode search experience to everyone in the US starting today. The chatbot-style addition to the company's search engine results page is designed to answer longer queries and uses Google's AI model to generate full responses based on--and linking back to--indexed websites on the open web. AI Mode is Google's direct response to the release of search engines from Silicon Valley startups like OpenAI and Perplexity, which provide chatbot-style answers to questions and queries. If all of this feels like déjà vu, that's because at last year's Google I/O developer conference, the company rolled out AI Mode's precursor, AI Overviews. In 2024, Google started to use its machine intelligence model to summarize the contents of the web and plaster a block of text at the top of the results for some queries.

It has been over two years since ChatGPT exploded onto the world stage and, while OpenAI has advanced it in many ways, there's still quite a few hurdles. Now, Austrian advocacy group Noyb has filed its second complaint against OpenAI for such hallucinations, naming a specific instance in which ChatGPT reportedly -- and wrongly -- stated that a Norwegian man was a murderer. To make matters, somehow, even worse, when this man asked ChatGPT what it knew about him, it reportedly stated that he was sentenced to 21 years in prison for killing two of his children and attempting to murder his third. The hallucination was also sprinkled with real information, including the number of children he had, their genders and the name of his home town. Noyb claims that this response put OpenAI in violation of GDPR.

A bill headed to Utah's Senate floor would require police to include disclaimers in any report written with help from artificial intelligence. Introduced by Sen. Stephanie Pitcher, SB180 comes nearly a year after multiple police agencies across the country began testing software like Axon's Draft One, prompting concerns from critics and privacy advocates. Draft One was announced by Axon in April 2024, kicking off a major new phase for the company best known for manufacturing tasers and a popular line of body cameras used by law enforcement. Axon built Draft One using Microsoft's Azure OpenAI platform, and is designed to auto-generate police reports using only an officer's body cam audio records. Once processed, Draft One then crafts "a draft narrative quickly," reportedly cutting down on police officer's paperwork by as much as an hour per day.

Artificial intelligence is expected to have an impact on the upcoming US election in November. States have been trying to protect against misinformation by passing laws that require political advertisements to disclose when they have used generative AI. Twenty states now have rules on the books, and according to new research, voters have a negative reaction to seeing those disclaimers. That seems like a pretty fair response: If a politician uses generative AI to mislead voters, then voters don't appreciate that. The study was conducted by New York University's Center on Technology Policy and first reported by The Washington Post.

Chat-based language models are designed to be helpful, yet they should not comply with every user request. While most existing work primarily focuses on refusal of "unsafe" queries, we posit that the scope of noncompliance should be broadened. We introduce a comprehensive taxonomy of contextual noncompliance describing when and how models should not comply with user requests. Our taxonomy spans a wide range of categories including incomplete, unsupported, indeterminate, and humanizing requests (in addition to unsafe requests). To test noncompliance capabilities of language models, we use this taxonomy to develop a new evaluation suite of 1000 noncompliance prompts. We find that most existing models show significantly high compliance rates in certain previously understudied categories with models like GPT-4 incorrectly complying with as many as 30% of requests. To address these gaps, we explore different training strategies using a synthetically-generated training set of requests and expected noncompliant responses. Our experiments demonstrate that while direct finetuning of instruction-tuned models can lead to both over-refusal and a decline in general capabilities, using parameter efficient methods like low rank adapters helps to strike a good balance between appropriate noncompliance and other capabilities.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Wisconsin Gov. Tony Evers on Thursday vetoed Republican proposals that would have allowed election observers to get closer to poll workers and required a new post-election audit, while signing into law a bill requiring that political TV ads using artificial intelligence come with a disclaimer. Evers, a Democrat, also signed a bipartisan bill exempting purchases of precious metal, such as gold and silver, from the state sales tax. The exemption does not apply to jewelry and other personal property, including works of art and scrap metal.