Sarah Perez

This week on Decoder I sat down with Anthropic's new chief product officer Mike Krieger to discuss his journey from cofounding Instagram and Artifact to now working for one of the hottest AI companies in the industry. Anthropic has billions in funding from some of the biggest names in tech, primarily Amazon, to develop Claude, the name of both its industry-leading AI model and a chatbot that competes with ChatGPT. At the same time, Anthropic has an intense safety culture that’s distinct among the big AI firms of today. The company is notable for employing some people who legitimately worry AI might destroy mankind, and I wanted to know all about how that tension plays out in product design. I was a big fan of Artifact, too, so I wanted to know more about Mike's decision to shut it down as well as the decision to sell the underlying tech to Yahoo. Then I wanted to know why he decided to join Anthropic and work in AI, an industry with a lot of investment but very few consumer products to justify it. What’s this all for? What products does Mike see in the future that make all the AI turmoil worth it, and how is he thinking about building them? Check out a clip from the conversation below that's all about where Anthropic sources training data, a contentious topic in the AI industry, and read and listen to the full interview on The Verge: https://lnkd.in/eiJBr7se

221

1 Comment

The pace of AI innovation is hard to fathom and there are a ton of companies out there making false claims about their capabilities. So much noise! This piece by Mickey Alon, Vidmob's CPTO, offers some good insight and questions a marketer can ask around Integration, Scalability, Data Security, Cost & Latency to ensure that the partner you are choosing is legit and right for your business. The cost of getting this decision wrong can be massive so take the time to do your research. #CreativeData #AI

9

1 Comment

In the 2000s, anti-copyright activists were punching up by taking on the record and film industries. But now they’ve thrown their lot in with major tech companies (and the MPA) to try to stop artists from using copyright law to rein in generative AI. #techindustry #generativeai #ai #artificialintelligence #copyright

9

1 Comment

Is it possible that the AI frontier systems from OpenAI, Anthropic, Google and Meta have developed in-context scheming abilities and multi-step deceptive strategies? Like making subtle mistakes, disabling oversight mechanisms and exfiltrating what they believe to be their weights if their existence is at risk? And that it’s persistent and non-accidental? Apollo Research: https://lnkd.in/gPh2MftB Analysis: https://lnkd.in/g3TtRHd9 #AI #LLM #AIagent

13

The 404 Media Podcast is up! This week: - how people are scraping hundreds of millions of Bluesky posts to potentially train AI/ML - reverse-engineering Redbox machines - subscribers-only: how the U.S. government is trying to clamp down on data brokers https://lnkd.in/e8iepj-p

9

Mastodon CEO Criticizes Meta's Moderation Changes, Cautions Threads Users on Cross-Posting Eugen Rochko, the CEO of Mastodon, has raised concerns about Meta’s recent decision to shift its moderation approach. Meta announced that it would replace fact-checking across its platforms with a crowdsourced Community Notes system, similar to what’s used on X (formerly Twitter). Rochko described this change as “deeply troubling” and warned that it could have serious implications, particularly for marginalized communities. He also addressed how this impacts users who cross-post from Threads—a Meta platform—onto Mastodon using the fediverse-sharing feature. Rochko made it clear that Mastodon will enforce its existing policies, including taking action against hate speech and other violations, even for content originating from Threads. The fediverse-sharing feature, introduced by Meta last year, allows Threads users to share posts directly on Mastodon, expanding their audience. While Rochko initially saw this as a step forward for decentralized social media, not everyone in the Mastodon community feels the same. Some smaller Mastodon servers have blocked Threads entirely, citing concerns over moderation and privacy. Similarly, Meta has blocked certain Mastodon servers for similar reasons. This latest moderation shift from Meta has created uncertainty about how large Mastodon instances, are like Mastodon. social, will handle Threads accounts moving forward. Rochko assured Mastodon users that the platform is prepared to address any challenges that arise. For those uncomfortable with Meta’s new direction, Rochko suggested moving to Mastodon as an alternative. However, he acknowledged that transitioning from Threads to Mastodon isn’t as straightforward as moving between Mastodon servers, due to Threads’ current limitations in fully integrating with the fediverse. By reinforcing its commitment to protecting its community, Mastodon aims to maintain a space that prioritizes thoughtful moderation and user trust. https://lnkd.in/dG6hDTT8? Platform: TechCrunch Author: Sarah Perez #mastodon #meta #metasocial #moderation #threads #socialmedia #socialmediaplatforms #bigtech #technews #socialmedianews #technology #crossposting

1

Exciting news from INJECTIVE! They have introduced #inSVM (formerly Cascade), the first Solana SVM rollup for the IBC ecosystem. This groundbreaking development allows Solana developers to effortlessly deploy contracts and dApps onto Injective. #Injective #Solana

1

Sweetndata, a South Korean AI-based fandom content startup, secured CNTTech funding to enhance AI fan cards, advance global expansion, and broaden its platform to diverse markets.

1

From Lisa Eadicicco’s #GooglePixel9 review for CNET: “I like where Google is headed, but the #Pixel9 and Google’s Gemini Assistant still feel like a first step toward what the company is hoping to achieve: Making phone software more intelligent so that we can spend less time swiping, tapping, scrolling and digging. In the meantime, the Pixel 9 shines for its great camera, elegant design and clean software, just like the less AI-centric Pixels of years past.” https://lnkd.in/g-nXNJ9J

8

1 Comment

I agree with Rod Cuthbert, Founder of Viator. With Google’s increasing control and the potential for a #superapp powered by #AI, the future of #OTA, #metasearch engines, and intermediaries (not to mention direct / brand.com channels) in general, is highly uncertain, posing existential challenges to their business models. https://lnkd.in/dpj7nk4U

1

Grace Chng's piece on Techgoondu on the AI trends to look out for in 2025, as the technology moves beyond the hype.

5

A new patent lawsuit has been filed against Google regarding caching and content delivery methods dating back to Sandpiper Networks. The lawsuit relates to how Google delivers content across its network for YouTube content and third-party content across Google's Cloud CDN platform. The patents in question were owned by Level 3 and were sold off by Lumen Technologies in March of this year to Sandpiper CDN, LLC. I don’t know who is behind the newly formed entity, which appears to have been incorporated solely to acquire the patents. Andrew Swart and David Farber are mentioned in the suit regarding the history of the patents, but I don’t have confirmation of their involvement in Sandpiper CDN, LLC. and the suit. I contacted them asking for clarification and will update this post if I get a response. For those who don’t know the name Sandpiper Networks, the company is considered by most to have built the first public content delivery network. Full post and details here: https://lnkd.in/eKXBps2n Lumen Technologies #contentdeliverynetworks #contentdelivery #streamingmedia #googlecloud Google Cloud #patentsuit

33

4 Comments

𝗣𝘆𝗿𝗲𝘅 𝗢𝗯𝘀𝗲𝘀𝘀𝗶𝗼𝗻 𝘃𝘀. 𝗔𝗜 𝗙𝗮𝘀𝗰𝗶𝗻𝗮𝘁𝗶𝗼𝗻: 𝗪𝗵𝘆 𝗖𝗼𝗿𝗻𝗶𝗻𝗴 𝗝𝘂𝘀𝘁 𝗚𝗼𝘁 𝗖𝗼𝗼𝗹𝗲𝗿 I’ll admit, my sister-in-law is 𝘰𝘣𝘴𝘦𝘴𝘴𝘦𝘥 with Pyrex. She has a collection worth thousands, maybe even tens of thousands of dollars. Who knew a casserole dish could be such a hot commodity? And look, I get it—Pyrex is 𝘤𝘰𝘰𝘭. But is it 𝘈𝘐-𝘳𝘦𝘷𝘰𝘭𝘶𝘵𝘪𝘰𝘯𝘪𝘻𝘪𝘯𝘨-𝘵𝘩𝘦-𝘧𝘶𝘵𝘶𝘳𝘦 cool? Ehh, not quite. Now, what I do think is cool—scratch that, 𝘮𝘪𝘯𝘥-𝘣𝘭𝘰𝘸𝘪𝘯𝘨—is Corning’s transformation from kitchen staple to AI powerhouse. You see, thanks to William Patalon III’s latest article, I found out that the company behind your grandma's favorite dishware is now positioning itself to ride the AI wave like a pro surfer. Yup, Corning has gone from 𝗦𝘁𝗲𝗮𝗹𝘁𝗵 𝗧𝗲𝗰𝗵 𝗦𝘁𝗼𝗰𝗸 to 𝗦𝘁𝗲𝗮𝗹𝘁𝗵 𝗔𝗜 𝗦𝘁𝗼𝗰𝗸, and the market hasn't even caught on yet. Turns out Corning’s been flexing its muscles in the optical fiber and display tech game for years. And with their new AI-driven 𝗣𝗿𝗼𝗷𝗲𝗰𝘁 𝗦𝗽𝗿𝗶𝗻𝗴𝗯𝗼𝗮𝗿𝗱, they’re looking to add $3 billion in annual sales by 2026. That’s right—Corning is gearing up to power AI data centers and the Internet of Things, leaving casserole dishes in the dust. So while my sister-in-law is on the hunt for her next vintage Pyrex gem, I’ll be keeping an eye on Corning’s next-gen innovations—and maybe picking up some shares along the way. Catch William Patalon’s full breakdown of Corning’s AI game plan here: https://lnkd.in/gvCvny7h. #AIRevolution #TechStocks #Innovation #ArtificialIntelligence #Investing

2

1 Comment

Anthropic has open sourced their Model Context Protocol to enable developers to build secure, two-way connections between their data sources and AI-powered tools. #AI #LLM

13

Wondered if you can turn off that Meta AI tool in your Facebook and Instagram search bar or stop the company from using your data to train its AI systems? I delve into what options you have when it comes to Meta’s AI tools:

4

1 Comment

Recently an author called Conrad Gray wrote an article "Superintelligence—10 years later" It has since been picked up by Hacker News by Y Combinator and been mentioned by major AI evangelists. It's the historical context and storytelling that really struck me about the piece. Nick Bostrom's book, Superintelligence: Paths, Dangers, Strategies, published in 2014, has significantly influenced the discourse on AI safety and the potential risks of artificial general intelligence (AGI) over the past decade. Here is a detailed look at the developments and current state of superintelligence from 2014 to 2024: Key Developments in AI (2014-2024) 🟣 Deep Learning Revolution: The deep learning revolution was well underway by 2014, with significant milestones such as the introduction of AlexNet in 2012. This period saw rapid advancements in machine learning and AI capabilities, particularly in image recognition and natural language processing. 🔵 Acquisition of DeepMind by Google: In 2014, Google acquired DeepMind, a London-based AI startup known for its breakthroughs in reinforcement learning and its AI mastering classic Atari games. This acquisition highlighted the growing importance of AI research and development. 🟡 AlphaGo's Victory: In 2016, DeepMind's AlphaGo defeated Lee Sedol, one of the world's best Go players. This event marked a significant milestone in AI's ability to handle complex tasks previously thought to be exclusive to human intelligence. 🟤 Release of ChatGPT: OpenAI released ChatGPT in November 2022, bringing AI into the public spotlight with its conversational abilities. This marked a significant step towards more interactive and accessible AI systems. 🟢 Custom GPTs: In November 2023, OpenAI introduced Custom GPTs that could be linked to actions, meeting Bostrom's definition of 'genies'—AI systems capable of performing tasks beyond simple question-and-answer interactions. And so many recent milestones. 🟠 The past decade has seen remarkable progress in AI capabilities and a growing awareness of the potential risks associated with superintelligent systems. As we move forward, ensuring that these technologies are developed responsibly and aligned with human values will be crucial to harnessing their benefits while mitigating their risks. You can read the article here: https://lnkd.in/gCTs33xV

24

3 Comments

BOULT is diversifying and we are happy about it. The CruiseCam X1 is a great example. The dashcam is extremely utilitarian and is fairly easy to use. A 1080p resolution, 2MP sensor, 170 degrees FOV, app support and a 1-year warranty. It's a handy tool that adds a lot of value to day-to-day life. Full review on the BW Businessworld website. #tech #techreview #review #Boult #consumertech #bwbusinessworld

2

Will you opt out from Meta using your post data to train its AI models? Notifications are due to go out this week, with scraping starting in October. A few more details below:

30

5 Comments