welcome to A.I. decoded, fast companyA weekly newsletter breaking the most important news in the world of AI. I’m Mark Sullivan. fast company, Covers emerging technologies, AI, and technology policy.
This week I focus on a new class of AI video generation tools that allow scammers to speak with completely different faces and voices in video calls. We also look at how Apple employees view Tim Cook’s appeasement of President Donald Trump and why OpenAI model writing has gone bad.
Sign up to receive this newsletter by email every week here. If you have comments about this issue or ideas for future issues, please contact us at: sullivan@fastcompany.comfollow us on X (formerly Twitter). @Sullivan.
scammer’s dream
Video generation models have improved over the years. We’ve always known that things could get weird when the generated video becomes almost indistinguishable from the actual video, and we’re now entering that stage. We are already seeing deepfakes of convincing political and sexual images. Another nefarious use of AI-generated video is in phishing scams.
Over the past six months or so, people have been posting to X about: new AI tools A feature that allows you to “swap faces” in real-time video. For example, AI might make it look like Leonardo DiCaprio or Scarlett Johansson are saying my words or acting out my facial expressions during a Zoom call. AI analyzes the user’s facial movements and vocalizations in real time, and transmits the message completely through the face of another person.
Some of these AI tools can generate new facial overlays using a single still image. So a fraudster anywhere in the world could find an image of a person on social media, use AI to create a reasonably convincing AI “face” from it, and then call one of that person’s relatives and ask for money. Scammers could use human voice samples to simulate the sounds of an AI overlay character.
click here Learn more about how AI face swapping plays a key role in deepfake fraud.
How Apple employees view Tim Cook’s relationship with Trump
Like other technology leaders, Apple CEO Tim Cook chose to engage with Donald Trump rather than ignore him. Mr. Cook has never publicly supported Mr. Trump or donated to his presidential campaign. But he has personally donated to Trump’s inaugural fund, visited the White House multiple times and is in at least semi-regular contact with the president.
During Trump’s first term, Cook spoke out against the Muslim travel ban, the transgender military ban, and the president’s “good people on both sides” response to the deadly white supremacist rally in Charlottesville, Virginia. By contrast, Mr. Cook has remained silent about the more extreme policies and actions that Mr. Trump brought to his second term. Until now. This week, Cook was forced to at least speak out against Apple employees.
Cook came under fire from employees and customers for attending a private VIP screening of Amazon’s new Melania Trump documentary at the White House on January 24, the same day that a federal Customs and Border Protection agent shot and killed Alex Preti, a nurse at a VA hospital in Minneapolis. This follows the January 7 shooting death of Minneapolis resident Renee Good by an Immigration and Customs Enforcement officer.
Mr. Cook remained silent about the Preti shooting until the evening of January 27th. reportedly Apple sent a memo to employees saying they were “heartbroken” by the deaths of Good and Preti.de-escalation” He also said he had discussed the matter with President Trump. It may have worked. On Wednesday, President Trump sought to calm the backlash against ICE. “We’re going to loosen up a little bit,” he said, echoing Cook’s words. (No actual de-escalation took place.)
click here I’d like to know more about how Apple employees are reacting to Cook’s relationship with Trump.
OpenAI sacrificed writing skills for GPT-5 coding skills
OpenAI’s latest model, GPT-5.2, has seen some backlash among users. People felt that the new model had less personality and less of its own opinion. And these shortcomings are reflected in its writing, which many users say is harder to read and sounds more unnatural than previous models.
OpenAI CEO Sam Altman admitted this week that GPT-5.2 is not as good a writer as its predecessor, GPT-4.5. “I think we messed it up,” Altman said. said At an internal “Developer Town Hall” on January 26th.
With so much money at stake, I don’t think it was just a “failure.” OpenAI had different priorities when developing GPT-5.2. “With 5.2, we’ve focused the most on making things really good in areas like intelligence, reasoning, coding, and engineering,” Altman said. “Bandwidth is limited here, so you can focus on one thing and ignore another.”
OpenAI also Industry-wide changes Chatbots are smart but dull, or chatbots that sound more like a responsible adult than an unconditionally supportive friend. There are safety and liability reasons for this. OpenAI is currently facing: multiple legal claims The company’s chatbot was unable to respond when vulnerable users (including minors) began talking about plans to take their own lives.
click here Learn more about OpenAI’s latest coding tools.
More information about AI from Fast Company:
Want exclusive reporting and trend analysis on technology, business innovation, the future of work, and design? sign up for fast company Premium.
