Facebook, TikTok and Reddit all updated their policies on misinformation this week, as Axios' Sara Fischer reports, suggesting that tech platforms are feeling increased pressure to stop manipulation attempts ahead of the 2020 elections.

Why it matters: This is the first time that several social media giants are taking a hard line specifically on banning deepfake content — typically video or audio that's manipulated using artificial intelligence (AI) or machine learning to intentionally deceive users.

• Deepfakes are videos (or audio) edited using AI and machine learning.
• Reddit said Thursday that it would ban accounts that impersonate individuals or entities in a misleading or deceptive manner. It will also ban deepfakes or other manipulated content that's "presented to mislead, or falsely attributed to an individual or entity."
• TikTok said Wednesday that it will ban misinformation that's created to cause harm to users or the larger public, including misinformation about elections or other civic processes, and manipulated content meant to cause harm. TikTok's policies do not explicitly address or define deepfakes, but they address manipulated content in much more depth than the company's previous standards.
• Facebook said Monday that it would ban deepfakes that were edited beyond adjustments for clarity and quality or were edited to misrepresent someone's statements.

The big picture: Concern around deepfakes began to surface after the 2016 election and has since become a popular talking point in accounts of our tech-fueled slide to dystopia.

• Yes, but: To date, there have been few instances of true deepfakes going viral to mislead users. Rather, most misleading viral media takes the form of more crudely doctored images and selectively edited video that don't use sophisticated technology but rather mislead by offering deceptive context.
• Case in point: Hazel Baker, Reuters' head of user-generated content news-gathering, told Axios last month that "Ninety percent of manipulated media we see online is real video taken out of context used to feed a different narrative."

Between the lines: The best example of confusion around whether a post was a deepfake and should be removed occurred last year, when a doctored video of Nancy Pelosi that was slowed to make her appear drunk went viral.

• Be smart: Facebook's new deepfake policies wouldn't necessarily ban that video, because it wasn't created using AI or machine learning. Reddit's new policies would, if the clip was posted with the intention to mislead users.

Our thought bubble: One of the biggest steps social media companies have made in taking action on deepfakes is defining what they are. Deciding when to remove them remains difficult.

• For now, the companies are trying to use intent as their barometer. But intent is highly subjective, and making those calls at tech-platform scale is going to prove a challenge.

Go deeper:

• TikTok expands content rules, cracks down on misinformation
• Facebook changing deepfake policies

Mark Zuckerberg said Thursday that he's giving up setting annual challenges for himself and trying to take a longer view. But 2020 has already thrown down a challenge for him: threading a needle between business demands and political landmines, as Sara reports.

The big picture: Zuckerberg has to grow revenue and users, yet not get blamed for tipping another election — and not buckle on what he views as the core value of free speech. Despite continuing criticism for Facebook and Zuckerberg in particular, he seems to be dodging any genuinely damaging blows — for now.

The reality: Facebook's revenue and user base have shown consistent growth over the past year, suggesting that users and advertisers aren't too spooked by the controversies around data privacy and misinformation. And unlike key rivals Google and Amazon, Facebook hasn’t really found itself in President Trump's crosshairs.

• Despite outrage from Democrats and some Republicans on Capitol Hill, there doesn't seem to be any meaningful regulation coming towards Facebook any time soon.
• And as the FTC's $5 billion Facebook privacy fine and $170 million YouTube child privacy fine last year show, big regulatory investigations into massive tech companies can deliver record penalties yet still not change the game.

Zuckerberg's plan is coming into focus:

1. Don't compromise on big, essential issues: On Thursday, Facebook said it wouldn't follow rival Google and limit micro-targeting on political ads, nor would it follow rival Snapchat and begin fact-checking them. Zuckerberg has said Facebook's permissive approach to political ads reflects a core commitment to free speech.
2. Make some cosmetic gestures: Instead, Facebook has made some small tweaks to its policy, like allowing users to opt out of political ads or certain micro-targeting options.
3. Engage with Trump and his campaign: To minimize White House backlash, Facebook has increased engagement with the Trump administration, including through a private dinner between Zuckerberg and the president last fall.
4. Better explain the logic: Zuckerberg has engaged in a lot more media outreach these days, sitting down with the press in the U.S. and overseas frequently. In December, he did a rare joint interview with his wife Priscilla Chan for CBS News. A memo leaked to the New York Times recently that lays out a trusted Zuckerberg lieutenant's perspective on the 2020 election contained some awkward passages, but also helped the company explain itself.
5. Buy favor: Facebook has jacked up its corporate ad spending significantly in the past year, especially in expensive messages targeted towards opinion leaders and policymakers. Facebook, along with its rivals, has used record lobbying dollars to try to purchase a Washington halo.

The bottom line: Facebook's public relations nightmare is far from over, but Zuckerberg is well on his way to meeting his 2020 challenge: compromise around the edges, don't buckle on things he cares about, and keep Wall Street happy.

Go deeper:

• Big Tech's Trump problem
• Facebook reports show Zuckerberg in charge, knuckles bared

A number of on-demand services companies are turning to layoffs to trim costs and shore up their balance sheets.

Why it matters: After years of seeing their unprofitable growth subsidized by plentiful venture dollars, companies are facing increased pressure to prove they can stand on their own two feet.

Driving the news: Here's a sampling of the cuts from just this past week:

• Scooter company Lime is exiting a dozen markets and laying off around 100 employees.
• Robot coffee shop Cafe X is closing three San Francisco outlets and cutting an unspecified number of jobs, as it focuses on its more high-volume airport locations.
• Rappi, a SoftBank-backed delivery startup operating in Latin America, laid off hundreds of employees, Axios has learned.

Netflix may have gone viral with a humorous tweet about password sharing this week, but Wall Street sees the practice as a drag on the company and its rivals to the tune of billions of dollars in revenue each year.

Why it matters: With competition growing and the cost of content continuing to increase, some streaming companies are likely to consider cracking down on the widespread practice.

Driving the news:

• As my colleague Felix Salmon noted, Netflix India tweeted a picture of a scam promising free Netflix and suggested followers just share passwords like everyone else.
• A new Parks Associates study estimates that password sharing costs streaming companies $9 billion last year, per Hollywood Reporter.

The big picture: In a way, Felix wrote, Netflix likes it when you use someone else's account. A paying customer is better than a non-paying customer, but a non-paying customer is better than no customer at all. And some password sharers can go on to become the next generation of paying customers.

• In the platform wars, size, growth and brand recognition matter more than anything else. Particularly in a fast-growing market like India, the prime goal for Netflix is to get as many people attached to its platform as possible — whether they're paying or not.

On Tap

• CES wraps up in Las Vegas, for those who haven't already made their escape.

Trading Places

• Former Visa executive Laura Barre Nadler is joining Afterpay as the chief financial officer for its U.S. business.
• David Gasca is leaving Twitter for a job at Google working on privacy for search and assistant.
• Former Tesla CFO Deepak Ahuja is joining Alphabet's Verily health unit as CFO, per CNBC.

ICYMI

• Sex-tech companies made the most of their time being fully included at CES (Wired)
• An under-the-radar company is helping law enforcement embed cameras in all sorts of places, including gravestones, car seats and vacuum cleaners. (Motherboard)
• Researchers found malware comes preinstalled on an Android phone offered by Assurance Wireless, a unit of Sprint that offers government-subsidized mobile service to low-income Americans. (Malwarebytes)

In case you could use a little good news.