TL;DR
Developments reported on Nov. 19 show AI becoming a workplace issue in union negotiations, software training, corporate investment and security operations. The evidence points to changing skill demands and new responsibilities, but it does not settle how AI affects jobs overall or how workers will share in its gains.
AI is becoming a workplace bargaining and skills issue across several developments reported Thursday, Nov. 19, from a union ballot at Rockstar Games to evidence on early-career software work and a major Anthropic financing. The reports show workers and employers confronting changes to job duties, training and security, while the eventual effects on employment and worker influence remain unsettled.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
More than 1,600 Rockstar UK employees are eligible to vote by Nov. 26 on whether to recognize the Rockstar Game Workers Union, part of the IWGB. The Register reported that the proposed agreement would cover negotiations on pay, overtime, pensions, bonuses and leave, and consultation rights on HR policies and AI. Rockstar publishes Grand Theft Auto VI, which began rolling out globally at midnight local time on Nov. 19. AFP reported the company has about a dozen studios and more than 4,700 employees.
In US software work, Indeed data show 4.5% of software development job postings were entry-level in the first quarter of 2026, compared with about 46% of postings across all occupations. Yet software development postings were up about 23% year over year in November 2026. Stanford researchers reported that employment among 22- to 25-year-olds in occupations more exposed to AI was 19% below a projected level based on less exposed peers. They described the results as early, descriptive indicators, not causal estimates.
A randomized Anthropic experiment involving 52 mostly junior engineers found that participants using an AI assistant to learn a Python library scored 50% on a follow-up quiz, compared with 67% among those who coded by hand. Participants who asked the assistant for explanations scored 65% or higher. Separately, Bloomberg and Reuters reported an Anthropic financing deal totaling $107.3 billion, including a $26 billion private placement. Its filing cited compute, infrastructure, and hiring and retention as uses for the funds.
AI is changing professional life through skills, bargaining and security
Recent developments show workers and employers facing new questions about job duties, training and workplace voice. The longer-term effects on employment and who shares in AI’s gains remain unsettled.
Workers seek a voice on AI
A union ballot could shape how Rockstar UK employees negotiate work terms and seek consultation on AI policies.
More than 1,600 employees can vote by Nov. 26 on recognizing the Rockstar Game Workers Union, part of the IWGB.
The ballot result is pending; the proposed agreement would cover employment terms and consultation rights on AI and HR policies.
Junior software work faces pressure
Hiring data and a small learning experiment raise questions for employers and training programs.
- Entry-level roles made up a smaller share of software postings than postings across all occupations.
- Software development postings still rose year over year.
- In a 52-engineer experiment, AI-assistant users scored 50% on a Python quiz, versus 67% for those coding by hand.
- Those who asked for explanations scored 65% or higher.
Investment brings new skill needs
Anthropic reported a $107.3 billion financing deal, including a $26 billion private placement.
Compute and energy
The filing cites compute and infrastructure needs, including energy projects requiring specialized expertise in short supply.
Hiring and retention
The filing also names hiring and retention; it does not establish how many jobs will result or how long demand will last.
Security work stays immediate
Software teams must respond to risks involving packages, credentials and system flaws.
- Two versions of Microsoft’s official Python Agent Framework package carried a credential stealer.
- Teams with affected versions should rebuild exposed environments and rotate accessible credentials.
- CISA urged organizations to patch Java4Shell; its record listed exploitation as “none.”
Workers Face New AI Bargaining Questions
The developments point to different kinds of workplace change. At Rockstar, a union vote could determine whether workers gain a formal channel to negotiate employment terms and seek consultation on AI-related policies. The ballot does not itself establish that the proposed terms will be adopted; its result is still pending.
For software teams, the figures suggest that entry-level hiring and training deserve attention even as demand for software development postings rises. The Anthropic experiment offers one limited result: participants who relied on AI for code performed less well on a quiz than those who practiced by hand, while explanation-seeking narrowed the gap. It does not establish how AI affects productivity or careers across the industry.
AI infrastructure investment also creates work beyond software. Anthropic’s filing says energy projects require specialized expertise that is in short supply. That signals demand for relevant skills, though the scale and duration of resulting jobs are not established by the filing. Meanwhile, package security and software flaws add operational duties for teams managing credentials and systems.
Products Worth Considering

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

Samsung 32 Odyssey G5 QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Game Development to Security Teams
The Rockstar ballot comes amid debate about the production of Grand Theft Auto VI, a game developed over more than eight years, according to AFP. Take-Two chief executive Strauss Zelnick said in February that generative AI had “zero part” in GTA VI. On Nov. 9, he told investors, “Hit creation remains the purview of human beings,” while saying recent advances had made it easier to create assets. Rockstar publishing head Jennifer Kolbe told Bloomberg, as quoted by The Register, “if they want a union, they will have a union.”
Two security incidents reported for Nov. 17 add another dimension to AI-related work. Versions 1.26.1 and 1.26.2 of Microsoft’s official Python Agent Framework package carried a credential stealer, according to the source material. PyPI quarantined the first version about 31 minutes after upload and the second eight minutes after upload. On the same day, CISA urged organizations to patch Java4Shell, a Java virtual machine flaw identified as CVE-2026-114938. CISA’s record listed exploitation as “none.”
The incidents do not show that AI caused either security problem. They do illustrate the maintenance and response work required when software components and credentials are part of organizational systems.
““Hit creation remains the purview of human beings.””
— Strauss Zelnick, Take-Two chief executive, speaking to investors on Nov. 9
Ballots and Job Effects Remain Open
The Rockstar ballot closes Nov. 26, and its result is not yet known. The full text of the proposed agreement and the tribunal judgment on 2025 dismissals were also not available in the source material. Zelnick’s comments describe Take-Two’s account of AI’s role; they do not independently establish every tool used throughout development.
The labor-market and training figures have limits. Stanford’s researchers cautioned that their findings are descriptive rather than causal, and the Anthropic experiment involved only 52 engineers learning one Python library. It is unclear how those results translate to other jobs, experience levels or longer-term career outcomes. The figures also do not establish that AI alone explains the share of entry-level postings.
Anthropic’s filing warns that rapid job displacement could prompt governments to slow AI adoption and says there is no assurance its own steps to ease displacement would work. The reported financing and infrastructure commitments do not settle how many jobs will be created, where they will be located or how long demand will last.
Votes, Patching and Hiring Decisions
The next immediate milestone is the Rockstar union ballot closing Nov. 26; the result will clarify whether eligible UK employees support union recognition. The proposed agreement’s terms and any related tribunal judgment remain matters to watch.
Employers and training programs will continue making decisions about entry-level hiring and how junior staff learn to review, test and debug AI-assisted work. Further labor-market data and research across more occupations will be needed to assess whether the patterns reported here persist.
Security teams have immediate response tasks. The GitHub Advisory Database advises treating environments that installed the affected package versions as exposed, rebuilding them and rotating credentials they could access, including CI and GitHub Actions secrets. Organizations also face CISA’s call to patch Java4Shell. The longer-term effects of AI investment, workplace bargaining and these operational demands are still developing.
Key Questions
What are Rockstar UK employees voting on?
More than 1,600 employees can vote by Nov. 26 on recognizing the Rockstar Game Workers Union, part of the IWGB. The proposed agreement would cover negotiations on employment terms and consultation rights on HR policies and AI.
Does the software jobs data show that AI caused fewer entry-level jobs?
No. Indeed reported that entry-level roles were 4.5% of US software development postings in the first quarter of 2026, while Stanford researchers described their employment finding as an early, descriptive indicator rather than a causal estimate. The data do not establish AI as the cause.
What did the Anthropic coding experiment find?
In a randomized experiment with 52 mostly junior engineers learning a Python library, those using an AI assistant scored 50% on a quiz, compared with 67% for those coding by hand. Participants who asked for explanations scored 65% or higher. The study was limited in size and scope.
What security issues were reported?
Two versions of Microsoft’s official Python Agent Framework package were reported to contain a credential stealer, and CISA urged organizations to patch Java4Shell. CISA’s record listed no exploitation for Java4Shell at the time described in the source material.
Source: Thorsten Meyer AI
