The Brief · Issue 9

Week of 24 August 2026 — record fines, record data volumes

Meta settles child-safety claims for up to $18B, Salesforce reports record Data Cloud volumes, and ISG says AI is moving analytics from reporting to action.

September 22, 2026

Compiled on 22 September 2026 for The Brief's launch archive, from sources published between 24 and 30 August 2026.

Four stories that mattered

Meta agrees to pay up to $18 billion over how it handled children's data

The Dutch case aside, this is a US state action: 52 attorneys general, led by California and Arizona, accused Meta of illegally collecting and using data from children under 13 and of designing Facebook and Instagram to be addictive to minors. The settlement running to $17 billion from 29 states plus a separate $1 billion from Texas requires default two-hour daily time limits for under-18 users, a midnight-to-6am night mode, non-algorithmic feeds by default for 13-17-year-olds, and new age-assurance measures within a year, all enforceable for a decade. The total rises to $18 billion if YouTube and TikTok adopt matching restrictions.

Analytics and growth teams at any consumer platform now have a concrete reference point for what "reasonable" data practices around minors look like, and for how expensive it is to get personalization and engagement metrics wrong when the users are children.

Source: ESG Dive, Meta settles landmark child privacy case for up to $18B

Salesforce reports record quarter as Agentforce and Data Cloud scale up

Salesforce's second-quarter fiscal 2027 results, covering the quarter ended 31 July 2026, showed total revenue of $11.3 billion, up 11% year over year (including roughly $456 million from Informatica). Combined Agentforce and Data 360 annual recurring revenue reached nearly $3.9 billion, up 210% year over year, and the company raised full-year revenue guidance to $46.1-$46.4 billion. Agentforce alone delivered 3.2 billion "agentic work units" in the quarter, a 97% increase from the prior quarter.

For analytics teams evaluating agentic tooling, the number to watch is less the revenue than the work-unit growth: it is one of the clearer public signals of actual usage, not just licensing, of AI agents inside a major enterprise platform.

Source: Salesforce, Salesforce Delivers Record Second Quarter Fiscal 2027 Results

AI helps scientists track Arctic permafrost thaw in near-real time

Researchers working through the Permafrost Discovery Gateway now actively monitor more than 4 million lakes across the Arctic, including roughly 70 million thermokarst ponds across the Alaska tundra, using AI to scan new satellite and aerial imagery for the appearance or disappearance of water bodies, a signal of permafrost thaw. The system combines aerial photography dating to the 1940s with satellite data from the 1970s onward; around 30 lakes on Alaska's Seward and Baldwin peninsulas drained in June and July 2026 alone.

It is a useful reminder that the hardest part of many AI-for-science projects is not the model but the decades of inconsistent historical imagery it has to be reconciled against.

Source: The Conversation, Disappearing lakes, old photos and AI are helping scientists map Arctic permafrost thaw in near-real time

ISG: AI is turning analytics software from reporting into action

ISG's 2026 Buyers Guide for Analytics, which evaluated 22 analytics software providers, found that AI and machine learning are now standard capabilities in business-intelligence tools and that agentic AI is starting to connect analysis directly to execution rather than stopping at a dashboard. Oracle led in five evaluation categories, Domo in three, and IBM in two; eight providers (AWS, Databricks, Domo, IBM, Oracle, Qlik, Salesforce and SAP) earned ISG's "Exemplary" rating. The guide weights product experience at 80% of the score and customer experience at 20%.

For buyers, the guide is a useful checkpoint on how far "agentic" claims have actually moved from marketing to shipped, ranked capability.

Source: ISG Research, Buyers Guide Analytics 2026 Executive Summary

Tool moves

One how-to

Before you trust a "zero-copy" data claim, ask three questions. Salesforce's headline this week was 104 trillion records ingested into Data 360, with zero-copy data sharing usage up 731% year over year. Zero-copy sharing — querying data in place, in someone else's storage, instead of duplicating it into your own warehouse — is genuinely useful, but vendors' usage numbers can flatter the technique. Before you rely on it for a production pipeline:

  1. Ask what "copy" means to the vendor. Some "zero-copy" integrations still materialize a cached or indexed version behind the scenes; you are avoiding a manual ETL step, not necessarily a second physical copy of the data.
  2. Check freshness guarantees, not just access. Querying a remote table live is not the same as querying a consistent snapshot — concurrent writes on the source system can change results mid-query unless the platform documents isolation behavior.
  3. Confirm who inherits governance. In a lakehouse or warehouse with native sharing, row- and column-level permissions on the source table need to propagate to the consumer, or you have created a quiet way to bypass access control.

None of this makes zero-copy sharing a bad idea — it usually beats another brittle nightly ETL job — but the pitch is "no copies, no delay, no drift," and it is worth verifying which of those three the vendor is actually promising versus implying.

One number

8 of 22. That is how many analytics software providers earned ISG's "Exemplary" rating in its 2026 Buyers Guide for Analytics — AWS, Databricks, Domo, IBM, Oracle, Qlik, Salesforce and SAP — out of the 22 providers ISG chose to evaluate. It is a useful shorthand for how concentrated ISG judges strong product-and-customer performance to be among vendors it already considers significant. It says nothing about the hundreds of analytics vendors ISG did not evaluate, and "Exemplary" is ISG's own weighted score (80% product experience, 20% customer experience), not an independent or standardized benchmark.

Source: ISG Research, Buyers Guide Analytics 2026 Executive Summary

Get the next one