Welcome to The Tech Fix, a weekly newsletter where we break down the latest technology news and trends from the advertising and marketing industry, curated by technology editor Jessica Heygate.

This is a big week for TikTok as it once again spars with the U.S. government over its Chinese ownership.

The Biden administration issued an ultimatum to TikTok parent ByteDance earlier this month to sell the app or face a possible ban in the U.S., one of its largest markets.

The order followed months of growing scrutiny over how TikTok shares data with its Chinese parent company. While TikTok has repeatedly denied that U.S. user data flows to China, several investigations suggest otherwise. In June, BuzzFeed found engineers in China routinely accessed U.S. data, while a former TikTok risk manager told Congress earlier this month that U.S. data may still be exposed under TikTok’s proposed plan to better lockdown the app.

TikTok executives have been presenting the plan, called Project Texas, at both public and closed-door events with major advertisers in recent weeks in an attempt to soothe concerns.

Historically tight-lipped about how its algorithm works, TikTok has also begun addressing concerns about whether content served to U.S. audiences and may be pushing certain agendas. A leader from the app’s data security division hosted a session at SXSW to provide details about how content is boosted on the app, a process which he said is controlled by a Los Angeles-based editorial team.

ByteDance has evaded pressure to carve out its operations before. Donald Trump’s administration pushed for a sale of TikTok’s U.S. operations to an American company in 2020. Microsoft considered a $30 billion acquisition of the U.S. business at the time, but ByteDance rejected the sale. It ended up reaching a compromise to sell a minority stake of TikTok to Oracle and Walmart, but this was shelved by the Biden administration.

TikTok CEO Shou Zi Chew is due to testify before Congress on Thursday. It will be his first grilling from the U.S. government, and TikTok employees are said to be nervous about his ability to convince lawmakers the app is not a threat. 

In the run up, TikTok has been attempting to rally support from its users. Chew posted a TikTok from Washington DC on Tuesday asking users to share why they love TikTok as he announced that the app has 150 million monthly active users in the U.S. Meanwhile, a group of TikTok influencers have gathered in DC on TikTok’s invitation to protest a potential ban.

Despite the regulatory concerns, Warc recently upgraded its ad spend forecast for TikTok to $15.2 billion in 2023, noting that 75% of marketers plan to increase their activity on the platform.

The AI explosion continues. Here’s a roundup of the latest products and developments:

• OpenAI said its new GPT-4 model was 40% more likely to produce factually correct responses than its predecessor, GPT-3.5, in internal testing. However, a report from NewsGuard found the newest model was more likely to surface false narratives in more convincing ways. NewsGuard said GPT-4 created responses mimicking Russian and Chinese state-run media outlets, health hoax peddlers and well-known conspiracy theorists.

• Microsoft has integrated OpenAI’s generative image generator, DALL-E, into Bing Chat to power an image creator, and is also integrating AI into its suite of Office 365 products, allowing users to control its features with text prompts rather than clicking through menu settings.

• A new generative AI chatbot has entered the fray: Chinese search giant Baidu unveiled Ernie Bot at a press conference last week, boasting similar capabilities to Microsoft’s Bing chatbot and Google’s Bard. Baidu demonstrated the chatbot’s capabilities via a prerecorded video instead of a live demo, which proved problematic for its peers.

• Adobe unveiled a family of AI models called Firefly that will enable users to generate images and content templates and edit videos across its suite of apps — including Express, Photoshop, Illustrator and Adobe Experience Manager — using text prompts.

• Adobe also announced a set of enterprise-focused AI services that will perform a range of marketing and sales tasks, such as automating ad copy, across its productivity apps.

• Mozilla launched a new startup called Mozilla.ai focused on building open source and “trustworthy” AI products. The nonprofit, which is behind the Firefox browser, has committed $30 million to the company, which it said will be “a space outside big tech and academia for like-minded founders, developers, scientists, product managers and builders to gather.” It said its initial focus will be to make generative AI safer and more transparent.

• TikTok will require users to label deepfake posts as part of an update to its community guidelines. Users who share images or videos that have been manipulated by AI to appear realistic, like a deepfake of Keanu Reeves, must label the content as “synthetic,” “fake,” “not real” or “altered” on a sticker or caption. The new guidelines go into effect on April 21. 

• More conservatives have accused ChatGPT of having a left-leaning bias, leading to calls for a right-wing alternative, The New York Times reports. This includes Elon Musk, who is apparently working on a less “woke” version of ChatGPT.

• A session at SXSW was created entirely by generative AI.

Consumers cite being concerned about sharing their data but their actions suggest otherwise, according to a new study from Razorfish.

Razorfish partnered with GWI to survey 1,600 U.S. respondents aged 16 and older about their attitudes towards data privacy. 

The vast majority said they were uncomfortable with their face scans being collected (78%), and with companies collecting photos of them (71%). But they don’t apply the same logic when sharing pictures of themselves on social media. A study from Statista in 2018 showed 62% of U.S. adults have uploaded a selfie to a social media website.

Users said they were most distrustful of social media and cryptocurrency companies. More than 56% said they “somewhat trust” or “trust a lot” healthcare providers and banks, even though these two industries record the highest number of cybersecurity incidents.

Pornhub parent company MindGeek was acquired by Ethical Capital Partners, a new Canadian private equity firm. The acquisition comes as MindGeek is embroiled in legal challenges over allegedly carrying sexually explicit videos of minors.

With new technology comes new lawsuits. Publishers are examining the extent to which their content has been used to train AI models, how they should be compensated and what their legal options are, The Wall Street Journal reports.

Keep up with Campaign! Drop us an email, comment on our stories, join our Facebook page, find us on LinkedIn and follow us on Twitter to be part of the conversation.