The AI Digest · Issue #3 · March 2026
March 13, 2026 · bindlcorp.com · 7 min read
Last issue we covered who’s winning and who isn’t in the AI economy. This week the story got more complicated. Sam Altman — the CEO of the company building the most widely used AI in the world — stood at a BlackRock summit and said out loud that some companies blaming layoffs on AI are lying. He called it AI washing. At the same time, real AI-driven cuts are happening at Atlassian and Block, Anthropic published research mapping exactly which jobs are most exposed, and GPT-5.4 crossed a benchmark that humans held for the first time. Here’s how to read it.
The Term Of The Week
Sam Altman Named It: “AI Washing.” Here’s What It Means And Why He Said It.
At BlackRock’s Infrastructure Summit in Washington this week, OpenAI CEO Sam Altman said what a lot of analysts have been thinking: some companies attributing layoffs to AI are using it as cover for cuts they were planning anyway. His framing was blunt — “Almost every company that does layoffs is blaming AI, whether or not it really is about AI.”
He wasn’t saying AI layoffs aren’t real. He was saying the label is being applied more broadly than the reality warrants — and that the distinction matters. When a company blames AI for a restructuring that’s actually about a post-COVID headcount correction or investor pressure, it muddies the data on what AI is actually doing to the labor market.
The numbers support his skepticism. The National Bureau of Economic Research surveyed thousands of C-suite executives across four countries this year. Nearly 90 percent said AI had no measurable impact on workplace employment in the past three years. Altman himself noted that fewer than one percent of 2025 job losses were directly attributable to AI. AI washing is real, widespread, and worth being skeptical about every time you see a headline.
~90%
C-suite execs who said AI had no measurable employment impact in the past 3 years (NBER, 2026)
<1%
Share of 2025 job losses directly attributed to AI — per Sam Altman himself
45K+
Tech layoffs globally by early March 2026 — AI the most frequently cited justification
What to do with this
When your company announces layoffs citing AI, it’s worth asking: is AI genuinely changing what work this team produces, or is this a restructuring decision that would have happened anyway? Those are different situations with different implications for everyone who stays.
If AI is genuinely driving the change, the roles that remain will look different — they’ll require fluency with the tools. If it’s AI washing, the restructuring reason is something else, and understanding what that is tells you more about the company’s trajectory than the AI framing does.
The Cuts
Atlassian Cut 10%. Block Cut 40%. The CEOs Were Unusually Honest About Why.
Atlassian cut 1,600 people this week — about 10 percent of its global workforce. CEO Mike Cannon-Brookes was specific in his framing: “It would be disingenuous to pretend AI doesn’t change the mix of skills we need or the number of roles required in certain areas.” The cuts free up capital to reinvest in AI and enterprise sales. Atlassian’s shares had already lost more than half their value since January as part of what traders are calling the SaaSpocalypse — a sustained selloff in enterprise software stocks driven by fears that AI agents will make conventional SaaS tools obsolete.
Block cut 40 percent of its staff — more than 4,000 people — the week prior. Jack Dorsey said the company is shifting to an “intelligence-native” model, with AI handling work that previously required large teams. Whether every one of those cuts is genuinely AI-driven is an open question. But both CEOs were more direct than most executives typically are, which makes these cases worth watching.
Atlassian
1,600
10% of global workforce. “AI changes the mix of skills we need.” Shares down 50%+ since January.
Block
4,000+
40% of total staff. Moving to “intelligence-native” model. Stock jumped the day after announcement.
What to do with this
The SaaSpocalypse is an investor story, but it has real implications for anyone working at, or building on top of, enterprise software platforms. If investors believe AI agents will make conventional SaaS obsolete, the companies building those tools are under pressure to prove otherwise — which means aggressive AI integration and cost-cutting. The companies that come out intact will be the ones that successfully pivot. The ones that don’t will continue to shed people.
If you work in SaaS or at a company heavily dependent on SaaS tools, pay attention to how your company is responding to this pressure. It’s the strategic question that will define which players survive the next 18 months.
The Research
Anthropic Built An Early Warning System For Job Exposure. Here’s What It Found.
Anthropic published research this week on which U.S. jobs are most exposed to AI — not in terms of which jobs will disappear, but which jobs have the highest percentage of their tasks that AI can already perform or assist with. The framing matters: exposure is not replacement. It’s a measure of how much of your job description is within reach of current AI tools.
The most exposed roles include paralegals, financial analysts, and software testers — white-collar jobs where a large share of daily tasks are codified, procedural, or research-based. These are also roles that are, statistically, held by people who are older, female, more educated, and higher paid — the opposite of the entry-level automation story most people expect.
The finding that matters most: jobs with high exposure are projected to grow more slowly through 2034. Not decline — grow more slowly. That’s a different kind of pressure than outright replacement. It means fewer new positions being created, more competition for the ones that exist, and a gradual shift in what those jobs require of the people doing them.
A note on what “exposed” actually means
High exposure doesn’t mean your job is gone. It means a meaningful share of your tasks can be done faster or replaced by current AI tools. The paralegal who knows which tasks to hand off to AI and which require human judgment is in a very different position than one who doesn’t. Exposure creates leverage for those who adapt. It creates vulnerability for those who don’t.
What to do with this
Look up your role specifically. Anthropic’s research is public and searchable. Understand what percentage of your tasks are in the high-exposure zone — and then figure out which of those tasks you should be doing with AI tools right now, before your employer or a competitor figures it out first.
The goal isn’t to replace yourself. It’s to be the person in your role who can do more because you’re using the tools — which makes you more valuable, not more replaceable.
This Week’s Move
Look Up Your Own Exposure Score. Then Do One Thing About It.
Anthropic’s research is public. The methodology is searchable. This week’s move is simple: find your job title in the data, understand what tasks in your role are considered high-exposure, and pick one of them to do with an AI tool this week — not as an experiment, but as your actual workflow.
It doesn’t have to be complicated. If your job involves research, use ChatGPT or Claude to do a first pass. If it involves writing, use it for a draft. If it involves summarizing data, feed it the data. The point isn’t to become an AI expert. The point is to move from “I know AI exists” to “I use AI as part of how I work” — which is the line that matters right now.
Sam Altman said this week that real displacement is coming, just not at the speed the headlines suggest. The window to adapt before it becomes urgent is open. It won’t stay open indefinitely.
Everything Else This Week
Quick Hits
GPT-5.4 crossed a human benchmark — OpenAI released GPT-5.4 this week, focused on coding and professional workflows. On OSWorld-Verified — a benchmark that simulates real desktop tasks — it scored 75 percent. The average human scored 72.4. The benchmark is narrow and specific, but it’s the first time a public AI model has crossed that line. OpenAI retired GPT-5.1 the same week, auto-migrating users to GPT-5.3 and 5.4. The pace of iteration is not slowing down.
“Too cheap to meter” — Altman’s phrase this week for where he thinks AI capacity is headed. The analogy is electricity: once infrastructure scales far enough, the marginal cost of intelligence approaches zero. If that happens, the companies that built competitive advantages on access to information or analytical capacity will need a new answer for what their edge actually is.
OpenAI crossed $25B in annualized revenue — and is reportedly taking early steps toward a public listing, potentially as soon as late 2026. Anthropic is approaching $19B. These are no longer startup numbers. The AI model market has become one of the fastest-growing sectors in tech in a span of about three years.
Zero-person startups — Altman described a wave of entrepreneurs, particularly in India, building companies that run almost entirely on AI agents: writing code, drafting legal documents, handling customer support, all from simple prompts. No employees. Compute instead of headcount. This is still early and niche — but it’s a preview of how startup economics are being rethought from the ground up.
BlackRock is spending $100M training tradespeople — The investment giant announced it’s funding training programs for plumbers, electricians, and HVAC technicians. Their CEO flagged a skilled trade shortage as AI reshapes white-collar work. The trades story — jobs that are hard to automate because they require physical presence and hands-on judgment — has been gaining traction. This is institutional money moving on that thesis.
New to AI tools?
The Window Is Open. Here’s How To Start.
Altman said this week that real AI displacement is coming — but not at the speed the headlines suggest. That gap between where things are now and where they’re going is the window. The workers who use it to build fluency will be in a different position than the ones who wait for urgency to force them into it.
If you’ve been putting off starting with AI tools, this week’s Anthropic data is the most concrete case yet for why that matters. Your job’s exposure score is measurable. The tasks that are in the high-exposure zone are identifiable. Starting now isn’t about anxiety — it’s about having options when the market catches up to where it’s heading.
Free guide — no account needed
Your First 20 Minutes With AI
Three tools. Three activities. 20 minutes. Built for people who know AI exists but haven’t figured out how to make it useful. No experience required.
Read the Beginner’s Guide →
BINDLCORP · AI Tools · Digital PDF
2026 AI Prompt Pack
$7.99 — Instant Download
90 fill-in-the-blank prompts across six categories —
zero technical knowledge required.
Copy a prompt, fill in the brackets, get a useful answer.
Works in Claude, ChatGPT, or any AI tool.
Includes 5 bonus tips, 5 advanced techniques, and a full shortcuts reference.
Comes in dark mode and light mode PDF.
42 pages, no blank pages, color-coded by section. Instant download.
Leave a Reply