Content Fans
Monday, May 18, 2026
All NewsAbout
Meta tests 'Hatch' AI agent for in-app shopping, summary featuresAI News & Trends

Meta tests 'Hatch' AI agent for in-app shopping, summary features

Meta is testing a new AI agent called 'Hatch' that may help with in-app shopping by summarizing chats, competitor offers, and drafting posts. Amazon appears to be using similar AI agents to speed up work tasks and fix problems more quickly in the workplace. Technology like ZaiNar's radio-wave timing might let AI agents find things in physical spaces, like warehouses. These trends suggest search and shopping may move from websites to being handled by AI inside apps, and that content may need to be made so agents can easily use and share it. Experts suggest this could change how people find and buy things, with agents guiding most of the process.

AI Infrastructure Costs Soar: Goldman Sachs Estimates $15-20 Million Per MWAI News & Trends

AI Infrastructure Costs Soar: Goldman Sachs Estimates $15-20 Million Per MW

AI infrastructure costs are rising quickly, with Goldman Sachs estimating new AI-ready data centers may cost $15-20 million per megawatt, much higher than traditional builds. This increase is partly due to higher power needs, advanced cooling, and more frequent hardware upgrades. Large companies like Microsoft, Amazon, Google, and Meta are expected to spend hundreds of billions on data centers and related technology in the next few years. Startups may struggle with these costs and often choose smaller models or shared resources. Experts suggest that finding ways to use resources more efficiently might become the main way companies compete as AI costs grow.

Newmark Secures $7.1B Loan for 1.2GW AI Data Center CampusAI News & Trends

Newmark Secures $7.1B Loan for 1.2GW AI Data Center Campus

Rising AI workloads may be causing developers to build larger data centers, with new financing structures needed to support these projects. In May 2025, Newmark reportedly arranged a $7.1 billion construction loan for Blue Owl, Crusoe, and Primary Digital Infrastructure to develop the second phase of a 1.2-gigawatt AI campus in Texas. Financing for such projects appears to focus on long-term power contracts and tenant pre-leases. Liquid cooling is becoming important for these data centers, as air cooling might not be enough for higher power densities. The movement of experts from advisory firms to owners suggests that in-house financial skills are becoming more important as the industry grows.

Anthropic's Claude uses 9-layer "burger" for AI context assemblyAI Deep Dives & Tutorials

Anthropic's Claude uses 9-layer "burger" for AI context assembly

Anthropic's Claude uses a nine-layer system, called the "context burger," to organize all the information it uses for each call. The layers start with important things like system prompts and environment data, and end with recent tool outputs and summaries, following a strict order. The research suggests that keeping context small and well-chosen may work better than giving the model too much information. When the context gets too large, Claude automatically trims less important data to stay efficient and focused. Engineers are advised to keep instructions clear and organized in the right layers to get the best results.

TikTok unveils MCP server to let AI agents run ad campaignsAI News & Trends

TikTok unveils MCP server to let AI agents run ad campaigns

TikTok has introduced the MCP server, which may let outside AI agents run everyday ad campaign tasks on its platform. This new tool appears to automate steps like setting up campaigns, changing budgets, and uploading creative content, possibly making these processes much faster. Some reports suggest advertisers could see faster performance testing and lower costs as a result. TikTok has not yet shared details on who can use MCP, and it may roll out access in phases. Other companies like Google and Meta also seem to be working on similar AI agent tools for their ad systems.

Latest News

EY withdraws loyalty report after GPTZero finds AI hallucinations
Business & Ethical AI8h ago

EY withdraws loyalty report after GPTZero finds AI hallucinations

EY withdrew a loyalty rewards report after the AI detector GPTZero found what appear to be made-up data, fake footnotes, and citations to sources that do not exist. EY is now reviewing how the report was published and says it is committed to using AI responsibly. The case highlights that using AI in public research may require stronger checks, like making sure each citation is real and having expert reviews before release. Some experts suggest that firms may need new rules to control how AI is used and to clearly tell readers when AI helped with a report. EY has not yet provided a timeline for finishing its review, and more changes to its process may follow.

Tariffs, Iran conflict raise US inflation, retail prices
Institutional Intelligence & Tribal Knowledge8h ago

Tariffs, Iran conflict raise US inflation, retail prices

Recent tariffs and conflict involving Iran may be raising costs for U.S. businesses and households, especially at stores and gas stations. Oil prices and shipping costs have gone up due to unrest in the Strait of Hormuz, and these increases appear to be affecting prices of everyday goods. Many companies are reacting by changing suppliers, moving production, or holding more inventory, which could add to costs. Food prices might climb faster than overall inflation, and experts suggest there is a risk of both higher prices and slower economic growth. Future changes in tariffs or energy markets might continue to add uncertainty for businesses and consumers.

TIME Expands Newsletter Growth With Simple Forward-to-a-Friend CTA
Institutional Intelligence & Tribal Knowledge8h ago

TIME Expands Newsletter Growth With Simple Forward-to-a-Friend CTA

TIME may be growing its newsletter by adding a simple line near the end that says, "If someone forwarded this to you, sign up here." This small message appears to make it easier for new people to subscribe and encourages current readers to share with friends. The strategy also includes social icons and a contact email, giving readers more ways to interact. Research suggests that having just one clear action at the end of the email helps increase clicks, and adding rewards for referrals might boost sign-ups even more. Tracking how readers respond to these options may help TIME understand what works best.

Taylor Swift, McConaughey Trademark Voices to Combat AI Deepfakes
Business & Ethical AI8h ago

Taylor Swift, McConaughey Trademark Voices to Combat AI Deepfakes

Taylor Swift and Matthew McConaughey are trying to use trademark law to protect their voices and images from being copied by AI deepfakes. They filed for trademarks because current laws may not cover new AI-made voice recordings. Experts say this is a new way to use trademark law, and it is not clear yet if it will work. Some new laws may help in the future, but they are not in place yet. This move may mostly help famous people with money, and it could shape how AI companies use celebrity voices and images.

Meta AI launches Incognito Chat for private WhatsApp conversations
AI News & Trends12h ago

Meta AI launches Incognito Chat for private WhatsApp conversations

Meta AI has launched Incognito Chat for WhatsApp, which may offer more privacy than other AI chat tools. Messages are processed in a secure space that Meta says even its engineers cannot access, and chats appear to disappear when users close the app or lock their phone. However, some metadata might still be kept for safety, and private chats are text-only for now. Experts warn that while the feature aims for strong privacy, there may still be risks if something goes wrong, and full invisibility cannot be promised.

Anthropic's Claude Code uses a 5-stage pipeline to compact context
AI Deep Dives & Tutorials12h ago

Anthropic's Claude Code uses a 5-stage pipeline to compact context

Claude Code appears to use a five-stage process to organize and compact information before sending it to its core language model. This process, sometimes described as a "context burger," stacks different types of information in a specific order, and most of the work may happen outside the model itself. Hierarchical instruction files, like short Markdown guides, seem to let engineers adjust the system's behavior quickly. Testing these prompts and using summaries instead of long histories might help teams save on costs and make their work faster. Some sources suggest treating this setup as flexible infrastructure, not just static text.

OpenAI Unveils Three New Realtime Voice Models for Enterprise AI in 2026
AI News & Trends12h ago

OpenAI Unveils Three New Realtime Voice Models for Enterprise AI in 2026

OpenAI has introduced three new voice models in 2026: Realtime-2, Realtime-Translate, and Realtime-Whisper, each designed for a specific audio task. These models may help businesses use voice AI in daily work by letting developers combine them as needed, instead of using one all-in-one solution. Reports suggest that Realtime-2 extends the memory for conversations and lets developers adjust how fast the answers come, which might keep call response times low. There are early signs of improved performance, but most results come from outside reports and are not confirmed by OpenAI. The new models appear to work well with current phone systems, but companies may need to balance better memory with costs and speed as they try them in real situations.

Anthropic's revenue run rate reaches $30B, potentially topping OpenAI's $25B
AI News & Trends12h ago

Anthropic's revenue run rate reaches $30B, potentially topping OpenAI's $25B

Recent reports suggest Anthropic's annual revenue run rate may have jumped to around $30 billion, possibly passing OpenAI's $25 billion run rate. Some sources indicate Anthropic might have closed the revenue gap with OpenAI, and there is strong market belief that Anthropic could have a higher valuation by the end of 2026. However, the way revenue is calculated for both companies can differ and may give a misleading picture of true financial health. Analysts say investors now care more about how much revenue companies keep and their profit margins, rather than just total sales. Both companies are spending a lot on infrastructure, and there is uncertainty about how they will be valued when they go public.

Meta tests AI agent "Hatch" for in-app shopping by mid-2026
AI News & Trends14h ago

Meta tests AI agent "Hatch" for in-app shopping by mid-2026

Meta is testing an AI agent called "Hatch" that may help users shop inside Instagram by mid-2026. Reports suggest Hatch could handle tasks like finding product details and checking out without leaving the app. Navigation apps and startups are also working on new ways to help people and vehicles move without always needing GPS. Marketing may shift as AI agents answer questions directly, so brands might focus more on being visible to these agents instead of traditional search results.

OpenAI integrates Plaid for ChatGPT Pro finance tools
AI News & Trends14h ago

OpenAI integrates Plaid for ChatGPT Pro finance tools

OpenAI has launched personal finance tools for US ChatGPT Pro users using Plaid, which may let users link their bank, credit card, and investment accounts securely. The feature is in preview and only available to some users for now. Users can reportedly ask questions about their spending and savings in plain language, and ChatGPT will answer based on real-time account data. OpenAI says the system is read-only and deletes data 30 days after disconnecting, but some security risks and privacy questions remain. The rollout is limited, and future changes or broader access may depend on user feedback and adding more account aggregators.