The Verge: All Posts by Adi Robertsonhttps://cdn.vox-cdn.com/community_logos/52801/VER_Logomark_32x32..png2024-02-27T15:00:48-05:00https://www.theverge.com/authors/adi-robertson/rss2024-02-27T15:00:48-05:002024-02-27T15:00:48-05:00Tumblr’s owner is striking deals with OpenAI and Midjourney for training data, says report
<figure>
<img alt="The Tumblr logo on a pink and purple background" src="https://cdn.vox-cdn.com/thumbor/m1Rt299VSylLl02uWbfuWmlgY7c=/0x0:2040x1360/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73168601/STK137_Tumblr_K_Radtke_01.0.jpg" />
<figcaption>Illustration: The Verge</figcaption>
</figure>
<p id="WRJG74">The owner of Tumblr and WordPress.com is in talks with AI companies Midjourney and OpenAI to provide training data scraped from users’ posts, <a href="https://www.404media.co/tumblr-and-wordpress-to-sell-users-data-to-train-ai-tools/">a report from <em>404 Media</em> alleges</a>. The report, based on an anonymous source inside the company, says that deals between Automattic and the two AI companies are “imminent.” It <a href="https://www.tumblr.com/jv/742956751128805376/what-is-this-about-the-tumblr-staff-wanting-to?source=share">follows nebulous rumors</a> that have spread on Tumblr over the past week, suggesting a deal with Midjourney could provide a new revenue stream for the site.</p>
<div class="c-float-left c-float-hang"><div id="xdXNJj"><div data-anthem-component="aside:12264153"></div></div></div>
<p id="QvIniw">According to <em>404</em>’s<em> </em>report, Automattic plans to launch a new setting Wednesday that will “allow users to opt-out of data sharing with third parties, including AI companies.” But it cites internal posts that suggest the company scraped an “initial data dump” containing “all...</p>
<p>
<a href="https://www.theverge.com/2024/2/27/24084884/tumblr-midjourney-openai-training-data-deal-report">Continue reading…</a>
</p>
https://www.theverge.com/2024/2/27/24084884/tumblr-midjourney-openai-training-data-deal-reportAdi Robertson2024-02-27T11:52:01-05:002024-02-27T11:52:01-05:00Texas sues Pornhub owner for not adding age verification requirements
<figure>
<img alt="Ken Paxton At Conservative Political Action Conference" src="https://cdn.vox-cdn.com/thumbor/6F7-l1vj5kqBKD4nZ8VIwhrQBMs=/0x0:6000x4000/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73168035/2025949801.0.jpg" />
</figure>
<p id="Rs6kBV">Texas Attorney General Ken Paxton has <a href="https://www.oag.state.tx.us/news/releases/attorney-general-ken-paxton-sues-major-pornography-distributor-violating-texas-age-verification-laws?ref=404media.co">sued the company behind Pornhub</a> for not complying with state-level age verification rules. <a href="https://www.oag.state.tx.us/sites/default/files/images/press/Texas%20v%20AYLO%20Filed%20Petition.pdf">The suit</a>, filed in Travis County District Court against Pornhub parent Aylo, accuses Aylo of failing to implement “reasonable age verification methods” on Pornhub and other adult sites that it owns. It seeks millions of dollars in damages under a law that was initially <a href="https://www.theverge.com/2023/8/31/23854369/texas-porn-age-verification-law-blocked-judge">blocked as unconstitutional</a>, then <a href="https://www.sacurrent.com/news/texas-age-verification-law-for-porn-sites-goes-into-effect-while-pending-appeal-33134494">allowed to take effect</a> last year.</p>
<p id="6igZW6">Texas is one of numerous states that have pushed for stricter age verification on either pornographic sites specifically or large parts of the internet as a whole. Its version of the law, <a href="https://legiscan.com/TX/text/HB1181/id/2819916/Texas-2023-HB1181-Enrolled.html">HB 1181</a>, requires that pornographic sites use either their own methods or a third-party...</p>
<p>
<a href="https://www.theverge.com/2024/2/27/24084615/texas-ken-paxton-pornhub-aylo-lawsuit-age-verification">Continue reading…</a>
</p>
https://www.theverge.com/2024/2/27/24084615/texas-ken-paxton-pornhub-aylo-lawsuit-age-verificationAdi Robertson2024-02-21T17:17:26-05:002024-02-21T17:17:26-05:00Google apologizes for ‘missing the mark’ after Gemini generated racially diverse Nazis
<figure>
<img alt="A Gemini image generation result featuring groups of almost entirely white men in white wigs" src="https://cdn.vox-cdn.com/thumbor/M5HGwJYj1WyjUg125i2QKunwgqQ=/0x0:1618x1079/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73155318/Screen_Shot_2024_02_21_at_2.56.19_PM.0.png" />
<figcaption><em>The results for “generate an image of the Founding Fathers,” as of February 21st.</em> | Screenshot: Adi Robertson / The Verge</figcaption>
</figure>
<p id="I04Lf1">Google has apologized for what it describes as “inaccuracies in some historical image generation depictions” with its Gemini AI tool, saying its attempts at creating a “wide range” of results missed the mark. The statement follows criticism that it depicted specific white figures (like the US Founding Fathers) or groups like <a href="https://twitter.com/JohnLu0x/status/1760170103474356519">Nazi-era German soldiers</a> as people of color, possibly as an overcorrection to <a href="https://www.vox.com/technology/23738987/racism-ai-automated-bias-discrimination-algorithm">long-standing racial bias problems</a> in AI.</p>
<p id="ZLNCT1">“We’re aware that Gemini is offering inaccuracies in some historical image generation depictions,” says the Google statement, <a href="https://twitter.com/Google_Comms/status/1760354549481546035">posted this afternoon on X</a>. “We’re working to improve these kinds of depictions immediately. Gemini’s AI image generation does generate a wide range of people. And that’s...</p>
<p>
<a href="https://www.theverge.com/2024/2/21/24079371/google-ai-gemini-generative-inaccurate-historical">Continue reading…</a>
</p>
https://www.theverge.com/2024/2/21/24079371/google-ai-gemini-generative-inaccurate-historicalAdi Robertson2024-02-16T09:30:00-05:002024-02-16T09:30:00-05:00You sound like a bot
<figure>
<img alt="3D illustration of a robot rubber stamping a text file." src="https://cdn.vox-cdn.com/thumbor/3tlzgTyYuGNZ_rF_bcA29vKaha4=/0x0:2040x1360/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73144155/246992_AI_at_Work_BORING_ECarter.0.jpg" />
<figcaption>Illustration by Erik Carter</figcaption>
</figure>
<p>AI used to be weird. Now ‘sounds like a bot’ is just shorthand for boring. </p>
<p>
<a href="https://www.theverge.com/24067999/ai-bot-chatgpt-chatbot-dungeon">Continue reading…</a>
</p>
https://www.theverge.com/24067999/ai-bot-chatgpt-chatbot-dungeonAdi Robertson2024-02-16T08:00:00-05:002024-02-16T08:00:00-05:00Why does Apple make it so hard to share the Vision Pro?
<figure>
<img alt="A man wears the Vision Pro, photographed slightly from behind" src="https://cdn.vox-cdn.com/thumbor/XjnomUV5JFrhiBNjT5jPD3IrB_c=/0x0:2700x1800/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73143841/246965_vision_pro_VPavic_0081.0.jpg" />
<figcaption>Photo by Vjeran Pavic / The Verge</figcaption>
</figure>
<p id="vOBQHg">Shortly before <em>The Verge</em> published its review of the Apple Vision Pro, I put it on to sit for some photos. The review unit had been fitted for our editor-in-chief Nilay Patel, but I’d worn it a few times as a guest and had a surprisingly good experience. That afternoon, though, I foolishly decided to skip the typical guest setup, which involves about a minute of calibration for the Vision Pro’s eye-tracking cameras. I put the thing on, and it didn’t work at all.</p>
<p id="bXjNyC">The Vision Pro’s cameras, I quickly realized, were expecting somebody else’s eyes. The cursor darted around wildly or refused to move. It wasn’t an unexpected outcome, but it drove home an inconvenient fact: not only would I need to go through the setup again, I’d need to do it...</p>
<p>
<a href="https://www.theverge.com/2024/2/16/24058318/apple-vision-pro-sharing-difficulties">Continue reading…</a>
</p>
https://www.theverge.com/2024/2/16/24058318/apple-vision-pro-sharing-difficultiesAdi Robertson2024-02-14T17:07:34-05:002024-02-14T17:07:34-05:00Terrorists are allegedly buying blue checks on X
<figure>
<img alt="An image showing the former Twitter logo with the X logo on its head" src="https://cdn.vox-cdn.com/thumbor/K8ZNV5eWegUqLLVpL6Edp_LP8lk=/0x0:3001x2001/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73140300/STK160_X_Twitter_006.0.jpg" />
<figcaption>Illustration: The Verge</figcaption>
</figure>
<p id="FP29m0">A report from the nonprofit Tech Transparency Project (TTP) alleges that X has been selling premium subscriptions to subjects of US sanctions, including leaders of the US-designated terrorist organization Hezbollah. The <a href="https://www.techtransparencyproject.org/articles/x-provides-premium-perks-to-hezbollah-other-us-sanctioned-groups">TTP report</a> identifies 28 accounts that were granted checkmarks under owner Elon Musk’s paid verification plan, evading rules that <a href="https://archive.ph/8qkE5">formally state</a> they’re banned from using it. The allegations raise new questions about how strictly social media platforms should vet users — after the Supreme Court ruled <a href="https://www.theverge.com/2023/5/18/23728423/supreme-court-section-230-gonzalez-google-twitter-taamneh-ruling">just last year</a> that the platform formerly known as Twitter was not responsible for abetting a terrorist attack.</p>
<p id="v9uzCg">The TTP report lists the full series of sanctioned entities that got verified on Twitter. The wide range of names...</p>
<p>
<a href="https://www.theverge.com/2024/2/14/24073146/x-twitter-blue-check-hezbollah-terrorist-groups-sanctions">Continue reading…</a>
</p>
https://www.theverge.com/2024/2/14/24073146/x-twitter-blue-check-hezbollah-terrorist-groups-sanctionsAdi Robertson2024-01-31T18:13:10-05:002024-01-31T18:13:10-05:00TikTok’s CEO can’t catch a break from xenophobia in Congress
<figure>
<img alt="The Senate Judiciary Committee hearing on online child sexual exploitation" src="https://cdn.vox-cdn.com/thumbor/90COQPdMAA1-fyj8t0lQRz39XS0=/2x0:5390x3592/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73101834/1968503944.0.jpg" />
<figcaption>Photo by Celal Gunes/Anadolu via Getty Images</figcaption>
</figure>
<p id="kZkwnk">Today’s <a href="https://www.theverge.com/2024/1/31/24056136/congress-child-safety-hearing-kosa-meta-x-discord-snap-tiktok">hearing on child safety</a> was — mostly — an unusually focused affair. The Senate Judiciary Committee called up the CEOs of X, Meta, Snap, TikTok, and Discord and grilled them for four hours on the potential dangers their services posed for children. Many of the lawmakers emphasized emotional impact, playing to an audience filled with families who’d had kids targeted by predators or otherwise harmed online.</p>
<p id="81lyXc">But midway through the hearing, it was dragged off course by a predictable tangent: the fact that TikTok is owned by Chinese company ByteDance. And a meeting ostensibly about keeping kids safe dipped into a <a href="https://www.theverge.com/2023/3/24/23654831/tiktok-congressional-hearing-xenophobia-china">now-familiar attempt</a> to make TikTok CEO Shou Zi Chew answer questions utterly unrelated to the rest of the day.</p>
<p id="0G8eB0">Although...</p>
<p>
<a href="https://www.theverge.com/2024/1/31/24057422/tiktok-ceo-child-safety-hearing-tom-cotton-xenophobia">Continue reading…</a>
</p>
https://www.theverge.com/2024/1/31/24057422/tiktok-ceo-child-safety-hearing-tom-cotton-xenophobiaAdi Robertson2024-01-31T12:06:00-05:002024-01-31T12:06:00-05:00All the news from Congress’ Big Tech child safety hearing
<figure>
<img alt="The US Capitol building during daytime" src="https://cdn.vox-cdn.com/thumbor/fw0Cj3jnuv7QX5zR9nLautLd_08=/194x0:1766x1048/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73100087/29947847713_6c70d75f1d_k.0.jpg" />
<figcaption>Image: <a class="ql-link" href="https://www.flickr.com/photos/uscapitol/29947847713/in/album-72157627522484962/" target="_blank">US Capitol / Flickr</a></figcaption>
</figure>
<p>The CEOs of five major tech companies will make a case for why their platforms are safe for kids.</p>
<p>
<a href="https://www.theverge.com/2024/1/31/24056136/congress-child-safety-hearing-kosa-meta-x-discord-snap-tiktok">Continue reading…</a>
</p>
https://www.theverge.com/2024/1/31/24056136/congress-child-safety-hearing-kosa-meta-x-discord-snap-tiktokAdi RobertsonLauren FeinerKevin NguyenAlex HeathMia Sato2024-01-31T08:30:00-05:002024-01-31T08:30:00-05:00The Vision Pro is a computer for the age of walled gardens
<figure>
<img alt="" src="https://cdn.vox-cdn.com/thumbor/1SU-UGuTA1choWscAgwwQ1R_piA=/0x0:2700x1800/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73099887/246965_vision_pro_VPavic_0265.0.jpg" />
<figcaption>Photo by Vjeran Pavic / The Verge</figcaption>
</figure>
<p id="gA5Et5">The Vision Pro, Apple’s new “spatial computing” headset, <a href="https://www.theverge.com/24054862/apple-vision-pro-review-vr-ar-headset-features-price">comes with a lot of limits</a>. It’s a technically impressive $3,499 device that’s straining against the basic capabilities of screens, cameras, eye tracking, and sheer component weight. Yet as I’ve watched the Vision Pro go from announcement to release, it’s also seemed held back by something that has little to do with hardware. Apple is trying to create the computer of the future, but it’s doing so under the tech company mindset of the present: one obsessed with consolidation, closed ecosystems, and treating platforms as a zero-sum game.</p>
<p id="fi9VJi">Apple is launching the Vision Pro with parts of the iPad catalog and a variety of specially tailored immersive content. But out of the box, you...</p>
<p>
<a href="https://www.theverge.com/24055677/apple-vision-pro-epic-netflix-app-ecosystem-monopoly">Continue reading…</a>
</p>
https://www.theverge.com/24055677/apple-vision-pro-epic-netflix-app-ecosystem-monopolyAdi Robertson2024-01-31T08:09:24-05:002024-01-31T08:09:24-05:00How to watch Linda Yaccarino, Mark Zuckerberg, and other tech CEOs testify in Congress
<figure>
<img alt="Mark Zuckerberg testifies in front of a microphone in a hearing in Congress" src="https://cdn.vox-cdn.com/thumbor/PZXyY7vDV7Ui1Y8JKYcJQd9qVoY=/0x0:6720x4480/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/73099864/1177753363.0.jpg" />
<figcaption>It’s not Zuckerberg’s first time in front of Congress. | Photo by Aurora Samperio/NurPhoto via Getty Images</figcaption>
</figure>
<p id="xiIJGl">Today, the US Senate Judiciary Committee will hear testimony from five CEOs of major tech companies: Linda Yaccarino of X, Shou Zi Chew of TikTok, Evan Spiegel of Snap, Mark Zuckerberg of Meta, and Jason Citron of Discord. The executives will answer questions on the topic of “Big Tech and the online child sexual exploitation crisis,” an ongoing issue for a Congress that’s introduced numerous child safety bills in recent years.</p>
<p id="3xTgoB">The hearing has been months in the making and apparently involved a little strong-arming from Congress, which reportedly <a href="https://www.theverge.com/2023/11/20/23969722/congress-subpoenad-linda-yaccarino-and-other-social-media-ceos-after-they-refused-to-cooperate">sent US Marshals</a> to subpoena Yaccarino and Citron. It begins at 10AM ET and will likely last several hours as lawmakers seize their opportunity to yell at some of both parties’ favorite bêtes...</p>
<p>
<a href="https://www.theverge.com/2024/1/31/24049456/yaccarino-zuckerberg-tiktok-x-meta-senate-judiciary-hearing-live-stream-how-to-watch">Continue reading…</a>
</p>
https://www.theverge.com/2024/1/31/24049456/yaccarino-zuckerberg-tiktok-x-meta-senate-judiciary-hearing-live-stream-how-to-watchAdi RobertsonLauren Feiner