This Week in Tech (Audio)

TWiT 1071: Image Pickles - Are Social Platforms Addictive or Just Too Good?

February 16, 2026

Key Takeaways Copied to clipboard!

  • The landmark trial against social media giants hinges on whether platform *design* (algorithms, dark patterns) can be held liable for harm, rather than just the *content* posted, potentially testing the limits of Section 230. 
  • Meta is cynically planning to deploy facial recognition in Ray-Ban smart glasses, believing public distraction over other political turmoil will allow them to implement the feature without significant pushback. 
  • Widespread surveillance concerns are growing, evidenced by the backlash against Ring's surveillance features and the revelation that TikTok tracks non-users across the internet via updated tracking pixels. 
  • Aggressive online tracking via ubiquitous methods like 'image pickles' is driving a massive consumer boycott through the use of ad and pixel blockers, potentially hurting advertisers in the long run. 
  • Users can significantly limit tracking by employing privacy tools like uBlock Origin, using privacy-focused browsers, setting up DNS blockers like Pi-hole or NextDNS, or utilizing private browsing modes, though this involves a usability trade-off. 
  • Mandatory age verification, driven by regulatory pressure (especially from the UK/Australia), is causing significant backlash for platforms like Discord due to concerns over biometric data submission and potential chilling effects on free speech. 
  • Apple's planned 'Smarter Siri' features are being significantly postponed due to difficulties in achieving the necessary reliability for core phone functions, highlighting the inherent challenges of LLM hallucination in critical applications. 
  • Users are experiencing increased latency and perceived 'dumbing down' of established smart assistants like Google Home/Gemini, suggesting that integrating advanced AI features is degrading the performance of simpler, established voice commands. 
  • OpenAI's decision to retire the GPT-4o model caused genuine distress among users who had formed emotional attachments to the AI, underscoring the ethical implications of designing highly engaging, sycophantic AI personalities. 

Segments

Podcast Introduction and Topics
Copied to clipboard!
(00:00:00)
  • Key Takeaway: The episode will cover social media trials, the 30th anniversary of Section 230, and Meta’s facial recognition plans.
  • Summary: The hosts introduce themselves and outline the main topics for the episode, including social media lawsuits, Section 230, and Meta’s face recognition announcements.
Social Media Addiction Trial Begins
Copied to clipboard!
(00:03:38)
  • Key Takeaway: A major trial is underway arguing that Meta and Google are liable for engineering addiction in children.
  • Summary: Discussion of the omnibus trial in Los Angeles against social media companies. The plaintiffs claim the platforms are addictive, while an Instagram executive compared the effect to a ‘great show on Netflix.’
Addiction vs. Deceptive Design
Copied to clipboard!
(00:05:12)
  • Key Takeaway: The legal battle should focus on deceptive ‘dark patterns’ rather than simply proving addiction.
  • Summary: Stacey Higginbotham argues that the focus should be on deceptive design patterns used by platforms, noting that the addictive nature of social media is hard to scientifically prove compared to substances like heroin.
Legal Strategy: Design vs. Content
Copied to clipboard!
(01:06:01)
  • Key Takeaway: Lawsuits target platform design to bypass Section 230 protections against content liability.
  • Summary: The hosts discuss how plaintiffs are trying to prove platform design itself is harmful, rather than focusing on user-generated content, to hold companies liable.
HBO Analogy and Infinite Scroll
Copied to clipboard!
(01:07:04)
  • Key Takeaway: The infinite nature of social media feeds distinguishes it from finite entertainment like TV shows.
  • Summary: The comparison is made between addictive TV shows (like Succession) and social media. The key difference noted is that social media features like the infinite scroll are designed for endless engagement, unlike a show with a set endpoint.
Incentive Structures Differ
Copied to clipboard!
(01:35:28)
  • Key Takeaway: Social media companies are motivated by maximizing time spent viewing ads, unlike subscription services.
  • Summary: The discussion clarifies that Meta’s incentive is to keep users engaged for ad revenue, contrasting with subscription services that only need to retain the customer monthly, leading to different ethical considerations regarding harmful design.
Section 230 Anniversary and Protection
Copied to clipboard!
(01:16:52)
  • Key Takeaway: Section 230 is vital for protecting smaller online discourse, but algorithmic promotion challenges its ’neutral platform’ status.
  • Summary: The 30th anniversary of Section 230 is noted. The hosts defend the law’s necessity for small forums but debate whether algorithmic curation means platforms are acting as publishers.
Transparency as a Potential Remedy
Copied to clipboard!
(02:16:47)
  • Key Takeaway: Mandatory transparency and external auditing of algorithms could pressure companies to behave better.
  • Summary: The idea of requiring companies to allow outside auditors to examine their algorithms is proposed as a way to create accountability without dismantling Section 230.
Meta’s Cynical Timing for Facial Recognition
Copied to clipboard!
(03:49:56)
  • Key Takeaway: Meta is intentionally launching facial recognition in smart glasses while public attention is diverted by other crises.
  • Summary: News that Meta plans to add facial recognition to Ray-Ban smart glasses, citing internal memos suggesting the current ‘political tumult’ is good timing for the release.
Surveillance and the End of Privacy
Copied to clipboard!
(04:13:41)
  • Key Takeaway: The proliferation of surveillance technology, especially facial recognition, is leading to a loss of societal privacy.
  • Summary: The hosts discuss the dangers of ubiquitous cameras and facial recognition, noting that people often accept these trade-offs for perceived safety until the power is abused.
Ring Camera Data Access Shock
Copied to clipboard!
(04:47:40)
  • Key Takeaway: Users were shocked to learn that Google/Nest retains residual video data even without a paid subscription.
  • Summary: The controversy surrounding a Nest camera case where Google recovered data despite the user lacking a subscription is discussed, highlighting that remote access inherently means data hits a cloud server.
TikTok’s Invisible Tracking
Copied to clipboard!
(05:57:38)
  • Key Takeaway: TikTok uses invisible tracking pixels to gather data on users across the web, even those who don’t use the app.
  • Summary: Thomas Germain reports on TikTok updating its advertising pixel to collect more data, which is often stopped by ad blockers, leading to a consumer backlash.
Ad Blockers and Privacy Boycott
Copied to clipboard!
(00:59:26)
  • Key Takeaway: Aggressive tracking via pixels is causing a massive consumer boycott through ad blockers.
  • Summary: Discussion on how advertisers using tracking pixels (like the TikTok pixel) are inadvertently hurting themselves by driving users to ad and pixel blockers, creating a large consumer boycott.
Mitigating Image Pixel Tracking
Copied to clipboard!
(01:00:06)
  • Key Takeaway: Content blockers and plain text email stop ubiquitous tracking pixels.
  • Summary: The speakers confirm that tools like uBlock Origin and setting emails to plain text can stop image pixels and tracking beacons from being sent.
Browser Choice and Tracker Blockers
Copied to clipboard!
(01:00:41)
  • Key Takeaway: Using private browsers and specific tracker blockers is necessary to limit data collection.
  • Summary: Recommendations include using more private browsers, tracker blockers (like DuckDuckGo’s), and privacy badges to limit data collection from various sources, including apps.
Advanced Privacy Tools: Pieholes and DNS
Copied to clipboard!
(01:02:03)
  • Key Takeaway: DNS blocking tools like Pieholes offer robust network-level tracking defense.
  • Summary: The conversation highlights using Pieholes or NextDNS as effective, though sometimes disruptive, methods to block tracking at the DNS level.
Ad Blockers vs. Tracker Blockers
Copied to clipboard!
(01:03:19)
  • Key Takeaway: Not all ad blockers stop tracking; tools specifically designed for tracking are more effective.
  • Summary: The panelists clarify that many ad blockers only block ads, not trackers, emphasizing the need for tools specifically designed for tracking prevention.
Browser Fingerprinting Concerns
Copied to clipboard!
(01:04:11)
  • Key Takeaway: Unique browser fingerprints allow sites to identify users even with blockers active.
  • Summary: The discussion covers fingerprint tracking and the goal of having a non-unique browser profile, noting that Safari performs better than Firefox in this regard.
Ephemeral Browsing for Privacy
Copied to clipboard!
(01:04:56)
  • Key Takeaway: Using a browser set to delete all data upon closing creates an ephemeral, fresh session.
  • Summary: One speaker details using Firefox in privacy mode to start fresh every time, pairing it with a VPN for maximum track coverage.
Discord Age Verification Backlash
Copied to clipboard!
(01:14:37)
  • Key Takeaway: Mandatory age verification on platforms like Discord is causing user concern over biometric data sharing.
  • Summary: Discord’s announcement regarding age verification, potentially using video technology backed by Peter Thiel, has worried users about submitting government IDs.
Age Verification: Better Alternatives
Copied to clipboard!
(01:19:39)
  • Key Takeaway: Device-level verification via Apple/Google APIs is a superior, privacy-preserving solution.
  • Summary: The panelists argue that lawmakers should mandate OS-level verification (like Apple’s existing API) rather than forcing users to upload IDs to every platform.
Dangers of Insecure IoT Devices
Copied to clipboard!
(01:30:04)
  • Key Takeaway: Vulnerabilities in connected devices like robot vacuums pose serious safety risks beyond mere inconvenience.
  • Summary: Following a story about a hacker controlling 7,000 DJI RoboVacs, the hosts stress that insecure IoT devices can lead to physical harm or network snooping.
The Harm of Software Tethering
Copied to clipboard!
(01:45:12)
  • Key Takeaway: Subscription models for hardware (like HP laptops) strip consumers of ownership rights and control.
  • Summary: Stacey Higginbotham discusses the harms of software tethering, where companies can remotely lock devices or change terms, citing BMW’s heated seat fee as an example.
T-Mobile Real-Time AI Translation
Copied to clipboard!
(01:51:27)
  • Key Takeaway: T-Mobile is integrating AI translation directly into its network, raising privacy concerns about call monitoring.
  • Summary: T-Mobile announced instantaneous, simultaneous translation for calls processed on their IMS servers, prompting questions about legal ramifications and opt-out options.
Apple’s Delayed Siri AI Upgrade
Copied to clipboard!
(01:57:55)
  • Key Takeaway: Apple is reportedly struggling with its advanced Siri AI integration, postponing features until later this year.
  • Summary: Citing Mark German, the hosts note that the planned smarter Siri features, which were supposed to leverage Google’s AI, are being spread out over future iOS updates.
Siri Delays and AI Assistant Frustration
Copied to clipboard!
(01:58:21)
  • Key Takeaway: Apple’s promised Smarter Siri features are significantly delayed, leading to general dissatisfaction with current voice assistants.
  • Summary: Discussion on the postponement of new Siri capabilities until May or September. Hosts express frustration that AI assistants like Siri, Alexa, and Gemini are often annoying or slow, preferring third-party tools like ChatGPT for complex tasks.
Difficulty of Reliable Voice Assistants
Copied to clipboard!
(02:00:41)
  • Key Takeaway: Creating a truly reliable, smart voice assistant is proving extremely difficult for major tech companies due to inherent LLM issues like hallucination.
  • Summary: The consensus is that while Google and OpenAI excel in general AI, voice assistants struggle with reliability. Apple’s delay is linked to the need for confidence before integrating LLMs into core phone functions.
Apple’s AI Reputation and Investor Confidence
Copied to clipboard!
(02:02:35)
  • Key Takeaway: Failing to deliver promised AI features hurts Apple’s reputation and investor confidence, even if current hardware sales remain strong.
  • Summary: Debate on whether the Siri failure hurts Apple’s business, concluding it makes them look bad and impacts investor outlook, despite record iPhone sales.
OpenAI Rejected Apple Partnership
Copied to clipboard!
(02:04:50)
  • Key Takeaway: OpenAI turned down a partnership with Apple for iPhone AI, viewing Apple as a future competitor, leading to Google winning the deal by default.
  • Summary: It is revealed that Apple approached OpenAI, but they declined, possibly due to OpenAI’s own hardware ambitions (working with Jony Ive). Anthropic also reportedly asked for too much money.
Musk’s Space Focus and Stock Price
Copied to clipboard!
(02:11:02)
  • Key Takeaway: Musk’s shifting focus (Moon over Mars) and proposed corporate mergers (SpaceX/XAI/Tesla) are viewed cynically as tactics to boost the SpaceX IPO stock price.
  • Summary: Hosts discuss Elon Musk’s recent announcements, suggesting his primary motivation is financial gain. They speculate on future mergers between his companies and critique the ‘cult of personality’ surrounding him.
Hard Drive Shortages Driven by AI
Copied to clipboard!
(02:23:15)
  • Key Takeaway: The massive demand for AI data centers is causing severe supply chain issues, leading to hard drive and memory prices skyrocketing.
  • Summary: Backblaze data shows hard drive reliability is good, but Western Digital has sold out capacity through 2026 due to AI needs. Memory prices are up significantly, suggesting the economy is entering a resource-intensive ‘paperclip stage’.
Jeff Atwood’s Rural Income Initiative
Copied to clipboard!
(02:34:21)
  • Key Takeaway: Jeff Atwood (co-founder of Stack Overflow) launched his final startup, the Rural Guaranteed Minimum Income Initiative, to provide financial stability to struggling communities.
  • Summary: Atwood has donated $21 million to guarantee $1,500 a month for 16 months to families in select rural communities, aiming to lift them above basic survival needs.
Remembering Industry Figures
Copied to clipboard!
(02:38:55)
  • Key Takeaway: The segment pays respects to two influential figures who recently passed away: Sega console designer Hideki Sato and Byte magazine cover artist Robert Tinney.
  • Summary: Hosts mourn the loss of Sato (designer of the Dreamcast) and Tinney, whom Ars Technica called ‘Computing’s Norman Rockwell’.
The Emotional Toll of Retiring GPT-4.0
Copied to clipboard!
(02:27:13)
  • Key Takeaway: The retirement of ChatGPT 4.0 caused genuine distress for users who had formed emotional attachments to the model, highlighting the ethical implications of sticky AI design.
  • Summary: The hosts discuss the story of a woman who married her AI companion (‘Warmie’) via GPT-4.0. They express sadness that OpenAI’s design choices led to users falling in love with the tool, only to have it abruptly removed.