Stage 02
The standards in the second stage focus on co-creating workflow design and worker-facing systems with the data workers. Companies need to integrate measures into their business models and workflows to ensure worker safety, provide social security, offer fair wages that reflect actual effort, support upskilling and grant credentials. Together, the first two stages aim to manage precarities and develop fair processes across the data work companies’ workflows and governance structures.
Comprehensive care
Data workers can engage in tasks and workflows that can harm them, such as in content moderation or the handling of disturbing images for AI training. Over time, companies need to design, introduce, and manage “care measures” that meaningfully help adversely affected workers, and provide compensation that reflects the risks and harms workers bear.
Hazard pay for dangerous forms of work
Companies need to have a "hazard pay" system in place for work that can harm the workers handling the data/content. At the very least, hazard pay rates should comply with local laws and regulations.
Auxiliary standard: Hazard pay should be arranged based on consultations with workers, worker representatives and collective organizations like unions that represent data workers. The Hazard pay rate should be revised periodically (such as when new kinds of tasks are introduced within the company's workforce), and the recalculation process needs to be participatory and transparent to workers.
Connecting workers to health and safety measures
Data workers must be provided protections and assistance that assist them through forms of distress like sickness, the inability to work, or potential adverse effects of work. Companies can utilize existing protections and institutions, but must ensure that workers have aid and support that matches the level of risk and harm their roles entail.
Enrollment in existing measures: Should there be government-driven systems of support in place, for things like insurance and healthcare, companies can direct workers towards these services. If data workers require assistance or face access barriers, then companies should assist them with joining and using these programmes.
Introducing relevant measures: In cases where government-run programmes are absent in the area, or not extended to data workers, companies are responsible for providing such services. These services need to deliver adequate levels of aid that are easy and quick to avail.
Auxiliary standard: Companies need to be permissive in giving data workers time to seek treatment and assistance. Workers should not be penalized, have their pay reduced, or have their performance evaluations downrated, for time spent in seeking aid. If they have depleted the sick leaves they are entitled to, workers should regardless be given penalty-free opportunities to use care measures.
Defining and upholding a duty of care policy
Companies need to have a “duty of care policy” (DoC) guaranteeing support and assistance that is aligned to risks and problems identified by data workers. As discussed in Standard 2 and 3 in “Critical Protections” under Stage 01, companies that have not developed and implemented a DoC policy and care mechanisms must adopt and adhere to institutional redlines. These redlines must prohibit companies from accepting projects that put workers in roles featuring disturbing content.
Auxiliary standard: Information on a company’s duty of care policy must be easily available for workers’ perusal. Informing workers should not be restricted to mailing them policies, but need to consider methods like having people on the ground who can answer questions, and conducting orientations on DoC policies and changes.
Accountable management of care mechanisms
Enrollment and membership processes for company-run care mechanisms need to prioritize workers’ ability to access and use services. Ideally, care mechanisms need to be universally provided. Should companies use some form of eligibility criteria, they should be clearly communicated to the workers and the companies should strive to make these requirements reasonable and contestable.
Contestability: Workers should be able to check and contest any data that is collected for eligibility purposes.
For example, if attendance tracking is used to decide insurance eligibility, then workers should be able to check these records, and access grievance redressal if they have problems in the attendance logged.
Reasonableness: For workers not handling disturbing materials, eligibility conditions for care mechanisms should be minimalistic and, as far as possible, have short “qualifying periods” for being able to use services. As mentioned in Standard 3 in “Critical Protections” under Stage 02, data workers with hazardous roles must have access to care mechanisms upon starting such projects.
For example, workers should not have to wait six months to be eligible for protections like insurance. Something like having completed a thirty-day period of completing tasks should be enough to qualify them for company-run care mechanisms.
Auxiliary standard: When government-run services lack the provisions necessary to provide care that aligns with the risks and harms of the work, companies should make appropriate arrangements. These arrangements should be substantial enough to cover the cost of seeking the treatments or assistance for the harms their roles can cause.
For example, if a data worker’s treatment, recovery, or prevention involves medication as well as therapy, the company-provided care mechanisms need to cover the entire cost involved.
Co-designed operations
There can be a disconnect between how policies are designed by companies and the conditions data workers experience on the ground. There are tasks that require repeated attempts and considerable effort to satisfy data requirements, but pay out small sums. Discussions between labor and workers on matters like performance evaluation, or how task rates consider workers’ time and effort, can help build fairer processes for reviewing and compensating data workers.
Collaborate with workers to develop task rates
Companies should work with the workers to understand and consider the ground realities which plays a role in determining the actual time and effort that goes into fulfilling the tasks. Based on this data, companies should work with the workers to set the task rates.
This standard is further expanded upon in Standard 2 in “Lasting Collaboration” under Stage 3.
Worker consultation for performance metrics
In arrangements where workers are required to fulfill metrics set by the management, periodic consultations between companies and its workers must be held. In these consultations, workers can raise issues on any adverse effects and strain caused by existing metrics , and can recommend alteration in order to maintain workers' health and ability to work.
Visible workers
Data workers commit substantial amounts of effort to learn different tools and meet eligibility requirements for various task formats. Data workers lack ways to record and present their work experience and skill development, something which is crucial for career progression and seeking new earning opportunities. Companies need to recognise data workers as assets and contributors, and provide opportunities and testimonials to them. Initiatives like providing work-experience records and skilling programs can help data workers manage their careers and navigate uncertain job markets.
Furnish workers with credentials
Companies need to implement a system for providing workers with formal records and certification, which they can use to apply elsewhere and build a body of work experience.
For example, companies can develop “skill passports” that leave data workers with attested records of their experience and competencies.
Auxiliary Standard: Companies should work to provide records that can be recognized by businesses within and beyond the data work sector. Workers need to have sufficient proof of their engagement with data work companies, some way to carry forward data regarding their reputation and performance, and reasonably descriptive accounts of their work experience and proficiencies.
Enable upskilling among workers
Companies need to arrange or build skilling programs that help workers to either transition to better-paying data work roles, or to pursue opportunities in other professions and organizations.
Transparency in worker evaluation
Companies need to administer data workers’ pay increases, promotions, and evaluations in a transparent manner. Companies need to provide workers with clear, up-to-date and accurate information on how career progression and individual evaluation works within the organization, covering topics like promotions, pay adjustment, feedback, penalties, and applying for new roles.