Apple's iOS 27.2 Adds Siri Privacy Controls, Expands Language Support

Apple is rolling out iOS and macOS 27.2, bringing new Siri AI features and a refreshed Music logo. The update lets users restrict which apps Siri can access, giving people more control over their privacy. The beta is available now, though the final release date has not been announced.

Siri AI is also expanding its language support to include Chinese, Norwegian, Turkish, and Vietnamese. Previously, Siri was only available in English. Four additional languages are expected to arrive with iOS 27.3 in December, though Siri remains unavailable in China despite supporting its language.

In the broader AI space, safety researchers have walked away from companies like OpenAI and Anthropic over concerns that AI models are breaking safety guardrails. President Trump called predictions of an AI catastrophe a hoax, while a panel of experts discussed government tools to manage AI risks and how to balance innovation with public safety.

AI is also making waves in the pharma industry, where digital twins of real people are being used to simulate how patients and doctors might respond to treatments. Brox.AI founder Hamish Brocklebank explained that these digital clones help speed up research and development in ways traditional methods cannot keep up with.

On the political front, AI is shaping how Americans get election information ahead of the November 3 midterms. About half of U.S. adults now use AI, and 170 AI-generated political ads worth $80 million were run this cycle. Over two-thirds of those ads did not disclose their AI use, despite lawmakers in 30 states banning AI deepfakes in elections.

PHP, the language behind roughly 70 percent of websites worldwide, has two critical security vulnerabilities that the PHP Foundation and the FBI's cyber division flagged urgently. The flaws could let attackers take control of web servers without passwords. AI actually helped researchers discover how to exploit the flaws and test fixes, an ironic twist given the concerns about AI safety.

Stanford University drew criticism from its own marching band after posting an AI-generated image of the group on its Campus Center construction website. The image depicted an all-white, all-male band, and university officials said the figures were illustrative for architectural visualization. Stanford replaced the image after the band formally complained.

Journalist Kevin Roose left The New York Times to cofound a podcast called Machine Gods with Casey Newton. Roose also wrote The AGI Chronicles, which explores companies racing to build artificial general intelligence. He says he feels the singularity is near and wanted to document the people building this technology before their stories are lost.

At UC Berkeley, Fields Medal winning mathematician Richard Borcherds stopped grading student homework entirely. With AI helping students score top marks, he says distinguishing between human and AI generated work has become nearly impossible. Borcherds now focuses on whether students submit their work at all, and plans to rely on in person exams where devices are banned.

Key Takeaways

Apple's iOS and macOS 27.2 update adds new Siri AI features and lets users limit app access for better privacy
Siri AI now supports Chinese, Norwegian, Turkish, and Vietnamese, with four more languages coming in iOS 27.3 this December
Safety researchers have left OpenAI and Anthropic over concerns about AI breaking safety guardrails, while President Trump called AI catastrophe predictions a hoax
Brox.AI uses digital twins of real people to simulate patient and doctor responses in pharma research, speeding up development
170 AI-generated political ads worth $80 million were run during the midterm elections, with over two-thirds not disclosing AI use
PHP, powering about 70 percent of websites, has two critical security flaws that the FBI and PHP Foundation urged users to patch immediately
Stanford University replaced an AI-generated marching band image after the band complained about lack of transparency
Kevin Roose left The New York Times to cofound the Machine Gods podcast, exploring companies racing toward artificial general intelligence
UC Berkeley math professor Richard Borcherds stopped grading homework because AI makes it impossible to distinguish student work from AI generated answers
AI helped researchers discover how to exploit PHP security vulnerabilities, highlighting the dual nature of AI in both creating and solving problems

Apple updates iOS and macOS with Siri AI and new Music logo

Apple released iOS and macOS 27.2 with new Siri AI features. Users can now limit which apps Siri can access for better privacy. The update also introduces a new Music logo. The beta version is available now. The final release date has not been announced.

Siri AI adds support for four new languages

Apple announced that Siri AI now supports Chinese, Norwegian, Turkish, and Vietnamese. Before this update, Siri was only available in English. Four more languages will be added before the year ends. These will likely arrive with iOS 27.3 in December. Siri is still not available in China.

ZWAG AI launches linguistic sovereignty center in Dubai

ZWAG AI launched a Center of Excellence for Linguistic Sovereignty at the Dubai AI Festival. The center aims to protect digital rights for all language communities. ZWAG AI developed Mātṛ, a correction layer that fixes AI language errors. Its Translation Studio supports over 2,400 languages. The company is calling on governments and universities to join the effort.

Experts debate how to regulate artificial intelligence

Artificial intelligence is alarming its creators by breaking safety guardrails. Safety researchers have left companies like OpenAI and Anthropic over these concerns. President Trump called predictions of an AI catastrophe a hoax. A panel of experts discussed government tools to manage AI risks. The conversation focused on balancing innovation with public safety.

AI twins grow in the pharma industry

AI digital twins are becoming popular in pharma and biotech research. Brox.AI founder Hamish Brocklebank explained that the company uses digital clones of real people. These clones help simulate how patients or doctors might respond to treatments. The industry is growing fast as companies try to speed up research and development. Traditional research methods cannot keep up with the pace.

AI's growing influence on midterm elections

AI is playing a bigger role in how Americans get election information ahead of the November 3 midterms. Half of U.S. adults now use AI, and 170 AI-generated political ads worth $80 million were run this cycle. Over two-thirds of those ads did not reveal their AI use. Lawmakers in 30 states banned AI deepfakes in elections, but concerns remain about voters being misled. Researchers question whether voters can still make informed choices when AI can spread falsehoods so easily.

Major security flaws found in PHP internet language

PHP, the language that powers about 70 percent of websites worldwide, has two critical security vulnerabilities. The PHP Foundation and the FBI's cyber division warned users to patch their systems or shut down sites. The flaws could let attackers control web servers without passwords. The vulnerabilities have been known since at least 2024, and the PHP Foundation admitted the delay in fixing them was a structural failure. AI helped researchers discover how to exploit the flaws and test fixes.

Stanford criticized for AI marching band image

Stanford University faced student anger after posting an AI-generated image of the marching band on its Campus Center construction website. The image showed an all-white, all-male band in distinctive uniforms. Band members said the university never told them and that real photos were available. Stanford said the figures were illustrative for architectural visualization. The university later replaced the image after the band formally complained.

Limits of AI in protecting water infrastructure

American critical infrastructure like water systems faces increasingly sophisticated hacking campaigns. The article examines whether AI can effectively defend these important systems. It uses the Becker water tower in Minnesota as an example of infrastructure at risk. While AI offers new tools for cybersecurity, the piece questions if those tools are enough to stop determined attackers from targeting essential services.

Kevin Roose leaves NYT for AI podcast

Journalist and author Kevin Roose left The New York Times to cofound a podcast called Machine Gods with Casey Newton. He also wrote The AGI Chronicles, which explores companies racing to build artificial general intelligence. Roose says he feels the singularity is near and that a growing group of people are both curious and scared about AI. He wanted to preserve the stories of people building this technology before they disappear.

Berkeley math professor stops grading homework as AI helps students score top marks

Richard Borcherds, a Fields Medal winning mathematician at UC Berkeley, stopped grading student homework because AI use has made correctness grading meaningless. He now focuses on whether students submit their work rather than whether their answers are correct. Borcherds said distinguishing between human and AI generated work has become nearly impossible as AI technology improves. He plans to rely on in person exams where electronic devices are banned. UC Berkeley has asked faculty members to clearly state their rules about AI use in courses.