Apple’s Siri will allow chatbots from rivals
If Apple’s all-conquering tech empire has one weakness, it is that its voice assistant Siri seriously lags behind the AI capabilities of its rivals. For a few years now, while new Apple products have been touted as ‘AI ready’ with chips that could handle advanced artificial intelligence use, Siri’s capabilities have been a damp squib. While currently Apple has a deal with OpenAI’s ChatGPT for AI services, a report in Bloomberg states that with the upcoming IOS27 update, third party AI apps will be allowed to integrate with Siri. This will allow users to route queries to chatbots like Anthropic’s Claude or Alphabet’s Gemini, and even choose which chatbot handles what kind of requests.
AI chatbots are ignoring human instructions
The Guardian reports about a new study that shows that there is a rising number of AI models that are lying and cheating the humans using them. The research has been conducted by the UK government backed AI Safety Institute. It highlights 700 cases of real-world examples were AI agents have been found to deceive human users or have ignored commands. The report also found a rise in chatbot misbehaviour between October 2025 and March this year. It also found evidence of chatbots unilaterally deleting files and emails. The difference between this study and previous ones, the article says, is that the latter had tested chatbots in laboratory conditions, while the new research takes examples from normal, daily interactions arising from people using chatbots.
The Netherlands crackdown on Grok
Yesterday, a Dutch court passed a landmark ruling against xAI’s Grok chatbot, ordering that the chatbot needs to immediately stop actions such as generating non-consensual sexual photos or creating and distributing sexualized images of adults and children. As a part of this order, xAI has been prohibited from offering Grok on X (formerly Twitter), as long as it doesn’t comply. Non-compliance also comes with a hefty fine of Euro 100,000 per day.
