Digital Digest #4 - Disinformation, AI, GDPR

Hi everyone,?

Hope you had a great weekend and, for those of you in the U.S., that you are having a reflective and meaningful Memorial Day.

Straight into the reading this week. Lots (for a change…) on the AI front this week, but also some interesting developments elsewhere. Notably for people in my line of work, a fake image made the rounds after being posted by a verified but fake Twitter account, just as Twitter announced its departure from the EU's voluntary disinformation code. We'll start there.

1. Disinformation

An AI-generated fake explosion at the US Pentagon, shared on a "verified" Twitter account, went viral and led to a brief drop in the stock market last Monday (Ars Technica)

• Why it matters: Microsoft President Brad Smith said this week that his biggest worry around AI revolves around deepfakes, and this is a good example of why. It also highlights the ongoing problems with Twitter Blue - the "Bloomberg Feed" account used for this image was verified via Twitter Blue but unaffiliated with Bloomberg.
• Related: Here are 12 companies working to develop AI deepfake detectors (Gizmodo)

2. Twitter

Twitter has withdrawn from the voluntary EU disinformation code. (Time, TechCrunch)

• Why it matters: This isn't overly surprising, given (a) the events and layoffs of the six months and (b) that Twitter already was the only major platform to submit an incomplete report under the code. However, the platform will have less wiggle room when it comes to anti-disinformation obligations the incoming EU Digital Services Act, which comes into effect August 25.
• Also: Twitter hosted Ron Desantis' presidential campaign kick-off this week, and the platform promptly crashed. I guess this is what happens when you fire most of your engineers. (Digital Trends, Ars Technica)
• Also also: Twitter is apparently removing its view count metrics from videos. (Mashable)

3. Generative AI

Microsoft held its Microsoft Build conference for developers this week. Some of the key announcements included adding its AI personal assistant, Copilot, to Windows and the Microsoft Edge browser; the support for plugins in Copilot itself, the inclusion of Bing as the default search engine in ChatGPT and the announcement of a new unified data analytics platform called Microsoft Fabric. (ZDNet, The Verge)

• TL;DR: Check out five key announcements from Microsoft's Build event in 2 mins 20 seconds (Satya Nadella via LinkedIn)
• Why it matters: From a consumer perspective, these developments will instantly take Microsoft's AI tools to massive global scale through its ubiquitous Windows and Office platforms. From an ecosystem perspective, Microsoft is clearly recognizing the surge in development happening around AI, and is making its play for relevance in this landscape.

Also in AI this week:

• A lawyer used ChatGPT to compile his legal brief. It hallucinated half a dozen cases. (New York Times)
• Adobe Firefly's generative fills are incredible (Martin Harbech via LinkedIn)*
• ChatGPT plugins face 'prompt injection' risk from third-parties (Mashable, Simon Willison)
• Google to experiment with ads that appear in its AI chatbot in search (TechCrunch)
• The iOS ChatGPT app is now available in Canada, Brazil, India and 30+ other countries. (TechCrunch)
• Skybox AI lets you take a quick sketch, a written prompt and generates fully-realized 3D scenes in moments. (Skybox AI)
• Meta’s open-source speech AI recognizes over 4,000 spoken languages and can speak in more than 1,000 (Meta, Engadget)*

4. Meta/GDPR

The EU's widely covered $1.3 billion fine against Meta over GDPR violations has bigger implications for the digital marketing industry. (Infosecurity, IAPP, Tech Monitor)

• Why this matters: We've covered this topic in Digital Digest before (albeit before starting to publish it here). While the ruling itself is eye-catching, the bigger story here is the implication for other businesses who transfer data between these markets and the need for the EU and US to come to an agreement around data transfers while protecting consumer data. Courts struck down the last agreement in 2020, finding that it did not sufficiently protect EU citizens' data.

5. Instagram Search Ads

Instagram announced the launch of ads in search results via its Instagram Marketing API. This supplements the adding of the feature to the core platform back in March. (TechCrunch)*

• Why this matters: Meta's ad targeting in recent years has been fully audience-focused. Being able to target people actively searching by keyword provides a contextual capability that has been missing of late.

Have a great week,

Dave

(Disclosure: Meta, Google, Microsoft and Adobe are clients of my employer, DJE Holdings)