AI's missing payoff, IT hiring's fine print, and why your bank won't save you
Apollo finds no AI payoff outside Big Tech, India's top five IT firms hire again but mostly freshers, Claude's incognito button, a bank-scam myth from Goa and Delhi, and Sweden's government switching staff to Copilot.
- Long Press · Daily Five26 Oct 202601 / 05
commentaryAI is paying Big Tech. Everyone else is still waiting.
Apollo chief economist Torsten Slok said in an October 2026 note that AI has delivered "zero financial payoff to corporate America outside of tech", pointing to flat revenue per worker and flat margins across the S&P 500 once you strip out its seven tech giants, as TechCrunch reported. I think he's right about where things stand, and it isn't much of a mystery. Most companies bought licences, ran a few pilots and left the actual work exactly as it was. That's a software bill, not a productivity plan.
The questionHas AI improved profits for companies outside Big Tech in 2026?
In one lineNot yet, according to Apollo chief economist Torsten Slok. In an October 2026 note he found flat revenue per worker and flat profit margins across the S&P 500 minus Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla, and said AI's financial payoff outside tech has been zero.
The problem
Companies have spent two years being told AI will make every worker more productive. The money flowing to Nvidia, Microsoft and the cloud giants is easy to see; the return for the companies paying them is much harder to find.
What changed
In a note published by Apollo, Torsten Slok looked at the S&P 493, the index without Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla. Revenue per worker and profit margins were flat, and he called AI "a great calculator" in a world where business intelligence is mostly judgment.
The catch
Flat averages can hide companies that are getting real gains, and productivity data usually lags a new technology by years. This is one economist reading aggregate numbers, not a company-by-company audit.
What it means for you
If your team's AI rollout is a chatbot licence and a training video, expect the same flat line. Pick one process, redesign it around the tool, and measure the hours it saves before buying more seats.
Questions people ask
What is the S&P 493 in Apollo's AI payoff analysis?
It's the S&P 500 with seven tech giants taken out: Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla. Torsten Slok used that group to check whether AI is helping ordinary companies. In his October 2026 note, its revenue per worker and profit margins were both flat, with no visible AI productivity boost.
Why does Apollo's Torsten Slok think AI isn't paying off for companies?
Slok calls AI a great calculator but argues that intelligence in business is mostly judgment. He says AI doesn't understand accountability, trust, culture, unwritten rules or fairness, so he doubts companies outside tech can turn cheaper intelligence into faster revenue growth and better margins. TechCrunch reported his note on October 25, 2026.
SourcesApollo ↗TechCrunch ↗
- Long Press · Daily Five26 Oct 202602 / 05
newsIndiaIndian IT is hiring again, mostly freshers and mostly for AI
India's five biggest IT services firms added 16,055 employees net from April to September 2026, up from 3,097 a year earlier, with TCS alone adding 13,537 and Wipro cutting 5,828, according to Angel One. It reads like a recovery and it isn't one yet. Freshers were 60 to 65 percent of the net adds, cheap to hire and quick to retrain, and the only demand that's actually growing is AI, cloud and security work. The jobs are coming back with a different job description.
The questionHow many employees did TCS, Infosys, Wipro and HCLTech add in H1 FY27?
In one lineIndia's top five IT firms added 16,055 people net between April and September 2026, against 3,097 a year earlier. TCS added 13,537, Infosys 3,948, HCLTech 3,158 and Tech Mahindra 1,240, while Wipro cut 5,828. Freshers made up 60 to 65 percent of the net additions.
The problem
Indian IT spent two years barely adding people while clients paused spending and AI started eating the routine coding and testing work that used to fill junior benches. Freshers and mid-level engineers have been left guessing whether the hiring will ever return.
What changed
In H1 FY27, TCS added 13,537, Infosys 3,948, HCLTech 3,158 and Tech Mahindra 1,240, while Wipro shrank by 5,828. Infosys alone added 4,480 in the September quarter to reach 332,542 people and still plans 20,000 fresher hires this year, per Deccan Herald.
The catch
Analysts don't call it a demand recovery. Overall IT hiring fell 3 percent while AI and cloud roles grew about 16 percent, and the full-year total may stay under 30,000 to 40,000.
What it means for you
If you're job hunting in IT, the door is open wider for freshers than for generalists with five years of maintenance work. Put AI, cloud or security work on your CV, even a small project, before the next campus cycle.
Questions people ask
Is Indian IT hiring recovering in 2026?
Only partly. The top five firms added 16,055 people in H1 FY27, five times last year's number, but UnearthInsights' Gaurav Vasu says it isn't proof of a broad demand recovery and expects fewer than 30,000 to 40,000 adds for the full year. Overall IT hiring fell 3 percent while AI and cloud roles grew.
How many freshers will TCS and Infosys hire in FY27?
TCS expects to hire 25,000 freshers this year. Infosys hired 8,000 in the first half and expects 20,000 for the full year, after adding 4,480 net employees in the September quarter to reach 332,542. Wipro has hired about 1,000 engineering graduates so far and plans to step up campus hiring.
Which IT skills are in demand in India right now?
AI, cloud, cybersecurity and specialised engineering. Demand for those skills grew about 16 percent year on year in the first half of FY27, according to figures reported by Angel One, while overall IT hiring fell 3 percent. That split matters more than the headline headcount number for anyone planning a career move.
SourcesAngel One ↗Deccan Herald ↗
- Long Press · Daily Five26 Oct 202603 / 05
tipClaude has an incognito button most people never notice
Claude lets you start an incognito chat by clicking the ghost icon at the top right of the screen, outside any project, and that chat stays out of your history and out of Claude's memory, according to Anthropic's help page. It's on every plan, Free included. I use it for the one-off stuff I don't want Claude bringing up again next week: a draft salary negotiation, a messy client email, a medical question. Just don't treat it like a shredder.
The questionHow do I start an incognito chat in Claude that isn't saved?
In one lineOutside a project, click the ghost icon in the top-right corner of Claude to open an incognito chat. It won't be saved to your chat history or Claude's memory, and Claude won't use it when searching past chats. It works on every plan, including Free.
The problem
Claude's memory and chat search are handy until something you pasted once keeps resurfacing in later answers, or sits in a history list on a shared laptop.
What changed
Click the ghost icon in the top-right corner (it only appears outside projects). The incognito chat isn't saved to your chat history or Claude's memory, and Claude won't pull from it when it searches previous conversations. Close it to go back to normal.
The catch
Anthropic still keeps incognito chats for at least 30 days for safety, and Team and Enterprise admins can see them in exports. It hides the chat from you and from Claude's memory, not from Anthropic.
What it means for you
Use incognito for one-off sensitive questions you don't want shaping future answers, and keep anything you can't share with Anthropic out of every chat, incognito or not.
Questions people ask
Are Claude incognito chats deleted immediately?
No. Anthropic's help page says incognito chats are retained for at least 30 days for safety purposes, and on Team and Enterprise plans they're included in data exports and follow the organisation's retention policy. Incognito keeps the chat out of your history and memory; it doesn't make it vanish from Anthropic's systems.
Can I use Claude incognito chat inside a project?
No. The ghost icon appears only outside a project, so you have to start a fresh chat from the main screen. Close the incognito chat when you're done and Claude goes back to saving conversations to your history and memory as usual. The feature is available on Free, Pro, Max, Team and Enterprise.
SourcesClaude Help Center ↗
- Long Press · Daily Five26 Oct 202604 / 05
myth busterIndiaYour bank is not your scam filter
Goa police on October 25 asked banks to hold senior citizens' large transfers to unfamiliar accounts until they're verified, after about Rs 100 crore was lost to cyber fraud in the state this year, as Daijiworld reported. That police have to ask tells you the check isn't happening by default. And sometimes the bank side is part of the problem: in Delhi a CSB Bank manager is among six arrested over 37 mule accounts. The safety net that works is a family rule, not a bank's fraud desk.
The questionWill my bank stop a digital arrest scam transfer in India?
In one lineDon't count on it. On October 25, 2026, Goa police had to ask banks to hold senior citizens' large transfers to unfamiliar accounts, after about Rs 100 crore was lost to cyber fraud in Goa this year. In Delhi, a CSB Bank manager is among six arrested over 37 mule accounts.
The problem
A lot of people, especially older account holders, assume a bank would stop or question a sudden transfer of their life savings to a stranger. Digital arrest and fake investment scams depend on that assumption.
What changed
Goa's cybercrime police asked banks to hold seniors' big transfers to unfamiliar accounts, flag sudden large deposits and reply to police faster. The same week, Delhi Police arrested six people including a CSB Bank manager over 37 mule accounts tied to 99 complaints and Rs 7.36 crore, per Delhi News Live.
The catch
Banks do run fraud checks, and most staff aren't involved in anything. But a request from police isn't a rule, and any hold has to work within RBI norms and the customer's right to their own money.
What it means for you
Agree one rule at home: nobody moves a large amount to a new account without calling a family member first. If money is going to several accounts, it's fraud. If it's already gone, call 1930 straight away.
Questions people ask
What did Goa police ask banks to do about cyber fraud?
At a meeting chaired by IGP Keshav Ram Chaurasia, Goa's cybercrime police asked banks to hold large transfers by senior citizens to unfamiliar accounts until verified, flag sudden big deposits, never share customer data, and answer police requests quickly. Police say Goa has lost about Rs 100 crore to cyber fraud in 2026, against Rs 101 crore last year.
What happened in the Delhi CSB Bank mule account case?
Delhi Police's North-West district team arrested six people, including a bank manager, over 37 mule accounts linked to CSB Bank's Saraswati Vihar and Tri Nagar branches. The accounts are tied to 99 cybercrime complaints across India and about Rs 7.36 crore. Police seized 63 Aadhaar cards and 60 PAN cards.
How can I tell if a money transfer request is a scam?
Goa police offered a simple test: if you're being told to send money to several accounts, it's fraud, so stop. Real police and agencies never collect money over a video call. If you've already paid, report it on 1930 or cybercrime.gov.in as fast as you can, since early reports give the best chance of freezing funds.
- Long Press · Daily Five26 Oct 202605 / 05
hot takeSweden picked the AI it can audit, not the smartest one
Sweden's Government Offices block public ChatGPT and Claude on staff computers from Monday, October 26, and point civil servants to Microsoft Copilot, the one tool they approved after a security review, as the Daily Northern reported from Swedish radio. I think it's the right call, and it says nothing about which model is smarter. Copilot sits inside a Microsoft contract and security setup the government already audits. For AI startups chasing public-sector deals, that's the real competition: paperwork, not benchmarks.
The questionWhy did Sweden's government block ChatGPT and Claude on work computers?
In one lineFrom October 26, 2026, Sweden's Government Offices block public ChatGPT and Claude on staff computers and direct civil servants to Microsoft Copilot, the tool that passed their security review. No detailed official reason was published, but the approval process, not model quality, decided it.
The problem
Civil servants handle sensitive material, and public chatbots make it easy to paste it into a service the employer has never vetted. Governments have struggled to say yes to AI without losing track of where that data goes.
What changed
From October 26, Sweden's Government Offices block the public ChatGPT and Claude services on work computers and direct staff to Copilot, which they reviewed and hardened, according to Yahoo News. Some AI systems stay available through controlled internal services, and Sweden's Tax Agency had already blocked AI services.
The catch
No detailed official reason has been published, and concentrating a whole government on one vendor's assistant carries its own risk if that vendor slips.
What it means for you
If your company hasn't approved an AI tool, assume the one you're using privately is the one that gets banned first. Ask IT which tool is approved before pasting anything confidential.
Questions people ask
Which AI tool can Swedish government staff still use?
Microsoft Copilot. Swedish media reported that the Government Offices reviewed Copilot, tightened its security and approved it, and staff are being directed to it instead of ChatGPT and Claude. Some other AI systems remain available through controlled internal services rather than the public websites.
Have other Swedish agencies blocked AI chatbots?
Yes. Before the Government Offices' block took effect on October 26, 2026, the Swedish Tax Agency and the National Board of Institutional Care had already blocked AI services, and the Enforcement Authority was reported to be considering the same step, according to Daily Northern's summary of Swedish public broadcaster reporting.
SourcesDaily Northern ↗Yahoo News ↗
Taps
Quick hits. One line each, for when a long press is too much.
- IndiaMediaNama ↗
Flipkart's super.money spent Rs 341 crore on marketing to earn Rs 292 crore
Its FY26 loss widened 78% to Rs 263.8 crore. Buying a top-five UPI spot is expensive, per MediaNama.
- IndiaInc42 ↗
Myntra slips into the red in FY26 with a Rs 582 crore loss
Flipkart's fashion arm is back to reporting a loss for the year, per Inc42.
- IndiaThe420 ↗
Surat police arrest two linked to 338 fraud complaints worth Rs 410 crore
Police allege the money was turned into USDT and sent abroad, with operators tied to Dubai and China, per The420.
- GlobalMacRumors ↗
Instagram launches an Apple TV app with Reels, longer videos and live shows
It needs tvOS 26 or later. Reels on the living-room TV is a bet on longer viewing sessions, per MacRumors.
- GlobalLA Times ↗
Elvis song-rights holders sue Anthropic over Claude training
The suit covers 274 songs; Anthropic disputes the claims, per the LA Times.