Work's Uncomfortable Summer: The Promises Are Cracking

Paper Cut took a week off for the summer. Welcome back. There's plenty to catch up on.

The headlines of the past fortnight tell a story that no one has put to a press release. A major tech company quietly shed 21,000 jobs over twelve months while presenting its AI strategy as an expansion. A government launched the first public tracker for AI-linked unemployment. A bank let staff work from home for the World Cup, weeks after demanding they return to the office. Senior workers at the peak of their careers are being asked, politely, to leave or to learn. The gap between what companies say about AI and what it is doing to their workforces has stopped being a talking point. It is now a data trail.

California published its AI Unemployment Tracker on 25 June, making it the first US state to formally attempt to quantify what technology is doing to work in real time. The tracker is still early and the data is directional rather than definitive. But the fact of its existence tells its own story. Policymakers no longer think the AI-and-jobs question is hypothetical.

FROM THE EBN NETWORK

Your employees know the truth. Does your EVP? At Fathom we measure the "Credibility Gap" between your promise and their reality.

Close the Gap

For employer brand teams, the next few months will require a specific kind of honesty. Trust in HR is reported as a competitive advantage by SHRM, but trust is not built through messaging. It accumulates through what organisations do when the numbers are down and the pressure is on. This fortnight, a lot of organisations revealed something about their character. The question for practitioners is whether their employer brand reflects that, or papers over it.

This Week’s Paper Cut's

Watch or listen here ↯

We keep tracking to a minimum. The video is hosted on YouTube and the podcast is hosted on PodBean, which may place cookies that measure viewing and personalize content. We only load the player after you consent to these cookies. You can update your cookie settings any time.

The consultancy model may have run its course

Accenture's difficulties this year have prompted something sharper than the usual market commentary. A Telegraph analysis published on 21 June described the consultancy model itself as structurally compromised, arguing that the combination of AI-capable enterprise software and reduced appetite for large-scale advisory mandates is compressing the case for big consulting engagements. Accenture has shed significant headcount over the past twelve months. The Financial Times separately reported that Anthropic has hired an economist with heterodox views on human-labour economics, suggesting that even the AI builders are stress-testing their assumptions about where value and human contribution sit.

The PwC 2026 AI Jobs Barometer, published this week, adds a more granular finding to this picture. PwC's research describes the global labour market as splitting into two distinct paths: one rewarding skills that AI cannot easily replicate, including complex communication, creative judgment, and human relationship management, and one compressing wages and headcount in roles where AI substitution is already underway. The research frames this as a skills premium story rather than a replacement story, but the trajectory is the same. Roles that can be automated are being automated, at different speeds in different sectors.

The SHRM data bookmarked this week noted that trust in HR functions with employees can itself be a competitive advantage. The specific claim is not a novel one, but the context matters. In a two-speed labour market where the pace of change is outpacing most workforce planning cycles, the organisations whose people functions have built credibility are better positioned to make difficult announcements and retain the people they most want to keep. Those who have not will face a more expensive problem.

Employer Brand POV: Consulting and knowledge-intensive service businesses are the canary for this fortnight's wider pattern. If your organisation competes for the same talent that consultancies are releasing, now is the time to audit your employer brand against what those candidates have just experienced. They arrive with expectations reset. Whether that is an advantage or a complication depends on what you actually have to offer.

Sources

Surveillance, RTO, and the battle for office legitimacy

The return-to-office argument has acquired two new awkward companions this fortnight. The first is the data on why it happens. A Wharton organisational psychologist, quoted in Inc, argued that RTO mandates are rarely about productivity, and that the real drivers are managerial visibility, commercial real-estate obligations, and the desire to preserve hierarchical proximity. A parallel New York Times opinion piece reached a similar conclusion. None of this is news to anyone who has sat in the relevant meetings, but peer-reviewed framing tends to travel further.

The second complication is surveillance. TD Bank in Canada has begun monitoring employee activity in the office, according to BNN Bloomberg. Meta paused an internal employee-tracking programme following an internal data leak, according to Wired. Canadian workplace law, BNN Bloomberg reported, offers workers limited protection against this kind of monitoring. The parallel stories are not coincidental. The logic of "if we're paying for the office, we want to know it's being used" is now manifesting as active monitoring infrastructure at scale.

Goldman Sachs and JPMorgan Chase provided the week's most legible illustration of the incoherence. Both banks, which have been among the most publicly committed to full-time office attendance, quietly allowed staff to work from home during the FIFA World Cup. The decision was operational, not principled. Which is perhaps the most honest thing either bank has communicated about remote work in three years.

The WSJ published a summary of the existing evidence on remote work and happiness this week, noting that the research is more complicated than either side of the debate acknowledges. A separate study reported in the Sydney Morning Herald found that remote work is associated with worse average mental health outcomes across the workforce, but significant improvements for specific groups including carers, people with disabilities, and those with long commutes. The aggregate hides considerable variance. Employer brand positioning on flexibility needs to be more specific than it usually is.

Employer Brand POV: Surveillance and inconsistent RTO policy are being read by employees and candidates as a signal about trust, not policy. Wharton researchers publishing findings about the non-productivity motivations for RTO mandates will surface in candidate research, review platforms, and journalist source requests. If your organisation has a surveillance programme or an RTO mandate you cannot articulate in terms of genuine business necessity, you have an employer brand liability that communications alone cannot resolve. The fix starts with policy, not messaging.

Sources

AI is creating a new kind of senior-talent problem

Two weeks ago, an account from a principal engineer who has spent three years inside AI-assisted software development was published on EBN. The piece described how AI tools compress certain high-skill tasks, creating a dynamic where senior expertise remains essential for design and oversight, but the visible labour of seniority, the coding, the review cycles, the artefact production, is increasingly handled by AI agents. The seniority signal is harder to read. The work looks the same from outside. The value is not.

That piece sits in direct dialogue with the broader signals this fortnight. A BBC report on Stanford's AI and the labour market found that elite university credentials still convey advantage, but that the advantage is narrowing faster at the mid-career level than at entry or senior levels. The white-collar baby boomer story in Business Insider, about experienced workers being asked to embrace AI tools or consider early retirement, fills in the other end of the age spectrum. MIT economist Simon Johnson, interviewed by the FT, offered the week's most quoted line: nobody needs as many white-collar workers as they used to.

Gen Z graduates, meanwhile, are framing their unemployment difficulties in terms of AI displacement, according to a Fortune analysis. The piece pushes back on this framing, arguing that macroeconomic conditions and the specific timing of their graduation cohort are doing more of the work than AI is. Both things may be true. The Gartner forecast, reported by Inc, that roughly half of AI job cuts would be reversed by 2027, is often cited as reassurance. What is less frequently noted is what Gartner means by "reversed": not that the roles return unchanged, but that the work is redistributed or restructured. That distinction matters for workforce planning.

HR Brew reported that HR leaders are now actively identifying and managing high-performing employees through AI transformation cycles with particular intentionality, treating human performance tracking as a counterweight to AI capability expansion.

Employer Brand POV: Senior talent is doing a quiet calculation right now. The question they are asking is whether your organisation is the kind of place that helps them evolve or one that is simply waiting for them to become dispensable. Roles emerging around AI oversight, model governance, and human-in-the-loop decision-making are the credible answer. If you are not hiring for, or visibly developing, those roles, you are not offering much to the most experienced people in your candidate pool. And they will notice.

Sources

AI adoption without accountability is producing a different kind of failure

A California Policy Lab tracker and new governor's announcement this week formalised something the data had been showing informally: AI's labour market effects are now significant enough to require dedicated public monitoring. The tracker, launched on 25 June, sits alongside a pattern of employer-level failures that point to a specific structural problem: organisations deploying AI tools before building governance, training, or accountability structures around them.

HRD reported this week that AI adoption without training is scaling mistakes rather than results. A parallel HRD piece noted that unsanctioned AI use across organisations is outpacing the guidance employers are providing. Less than 5% of companies, according to HRD citing an external report, are achieving what researchers categorise as transformational outcomes from AI. The gap between AI investment and AI results is persistent.

The AI burnout piece from Inc was among the most-shared this fortnight. It describes a specific phenomenon: employees who are deploying AI tools without any accompanying redesign of their workload expectations or performance benchmarks, and who are experiencing cognitive overload as a consequence. Managers are reported to be largely unprepared for this. The Japan Today survey on skills people outperform AI on, including empathy, creative synthesis, and moral reasoning, sets up a coherent implication: if those skills are where human advantage lies, organisations need to be explicitly cultivating them, not simply deploying AI on top of unchanged job descriptions.

KPMG's decision to roll back its Summer Friday half-day arrangement, reported by CityAM, was framed as a cost-efficiency signal. In the context of the fortnight's broader pattern, it reads as something more pointed: a professional services firm removing a visible, low-cost flexibility benefit at a moment when its competitors are shedding staff and its clients are rethinking what they need advisory firms for. The optics are not accidental. They are the product of a decision.

Employer Brand POV: Deploying AI without redesigning the expectations around it is not a technology problem. It is a culture and governance problem, and it is beginning to show up in burnout data, unsanctioned tool usage, and employee sentiment. Before your next AI rollout, the question for HR and EB is not "how do we communicate this" but "have we changed the job?" If the answer is no, the communication problem will solve itself, badly, via Glassdoor.

Sources

The Cut List

Workforce reductions reported during the fortnight of 12 to 26 June 2026.

Oracle: Shed approximately 21,000 roles over the preceding twelve months as part of an AI-focused restructuring, according to WSJ and BBC reports published this week. Oracle confirmed the scale of job eliminations in response to reporting.

Lucid Motors: Announced plans to cut approximately 18% of its US workforce, with COO departing simultaneously. Reuters, 22 June 2026.

Shopee: Laid off staff globally, with Singapore cuts reported to have hit product and engineering teams. Vulcan Post, 26 June 2026.

Diageo: Preparing deep job cuts amid a global demand slowdown in the drinks sector. Vino Joy News, 26 June 2026.

Lazada: Joined a growing list of firms reducing headcount in Singapore. Straits Times, 26 June 2026.

Bungie (SIE): Layoffs linked to performance struggles with its Destiny franchise. Inven Global, 26 June 2026.

Singapore broader picture: Retrenchments rose in Q1 2026, with degree holders and older workers disproportionately affected, according to CNA and Vulcan Post reports citing Ministry of Manpower data.

Australia: Headline unemployment eased but the trend rate reached its highest point in four years, according to ABC News reporting on ABS data released 25 June 2026.

Sources

Closing thoughts

The pattern across the past fortnight is not subtle. Organisations are announcing AI transformation while cutting staff, demanding office attendance while making exceptions for football, and deploying tools without the governance structures that would make them safe or effective. The California AI tracker, the Wharton research on RTO motivations, and the data on AI burnout and unsanctioned tool use all point to the same gap: the story being told externally and the reality being experienced internally are diverging.

That divergence is what employer brand practitioners are being paid to close, or at minimum to be honest about. The fortnight's signals suggest the window for vague reassurances is narrowing.

Takeaways

• The PwC 2026 AI Jobs Barometer describes the global labour market as splitting into two paths: one rewarding durable human skills, the other compressing roles subject to AI substitution, with the speed of change varying by sector.
• California launched the first public AI unemployment tracker on 25 June 2026, marking a shift from AI-and-jobs as policy speculation to AI-and-jobs as a formally monitored data question.
• Less than 5% of companies are achieving transformational outcomes from AI deployment, and unsanctioned AI use by employees is now outpacing official guidance.
• Meta paused an employee-tracking programme following a data breach, and TD Bank began office surveillance without clear legal protections for workers, illustrating the speed at which monitoring infrastructure is outpacing policy frameworks.
• SHRM data indicates that trust in HR functions can be a competitive advantage in tight labour markets, but trust is built through decisions and consistency, not communications programmes.
• Simon Johnson, MIT economist, stated that no organisation currently needs as many white-collar workers as it used to, and Gartner's forecast that half of AI job cuts will be "reversed" by 2027 refers to role redistribution, not role restoration.
• The pattern of RTO mandates combined with flexible exceptions for events like the World Cup is being read by employees as evidence that the productivity rationale is a post-hoc justification rather than the actual basis for return-to-office policy.
• Gen Z unemployment in the current cycle is more attributable to macroeconomic conditions and graduation timing than to AI displacement directly, but the subjective experience of AI competing for entry-level roles is shaping candidate attitudes toward employers that communicate poorly about technology and people strategy.

Trending on EBN