“I like that the deputy CIO’s team provides standards and does testing, and that they bring in our community to determine usability,” Zuback says. “It might pass the standards, but is it usable? This is greatly appreciated by our community.”
“This is not just about checking a box,” Kron says. “We are aiming for usability, not just accessibility.”
The NSA’s accessibility standards are the same as those of other agencies in their overall alignment to Section 508 of the Rehabilitation Act of 1973. The agency reinforces Section 508-specific contract language in its technology acquisition processes. All NSA websites and applications must meet the NSA’s Information and Communication Technology Accessibility Standards (excluding certain missions or imported collateral) using the NSA ICT Accessibility Standards scoresheet. Whether through its own development or by testing third-party products, the NSA maintains a scoresheet that rates accessibility, using a scale of 1 to 5 to assess whether a project or product is fundamentally inaccessible for all persons with disabilities (1) or functionally accessible (5).
DISCOVER: Technology is adapting to disabled federal workers.
Using AI to Improve Digital Accessibility
Nearly 60 percent of federal websites are not fully usable by those who require assistive technology, according to comments from Federal CIO Clare Martorana regarding the Office of Management and Budget’s memorandum M-23-22, “Delivering a Digital-First Public Experience.” The issue of digital accessibility, and employees’ ability to access both external and internal systems and intranets, is a particular challenge, not just within the IC but within all federal agencies, Martorana notes.
Generative AI represents just one rapidly developing resource that could address accessibility issues on digital systems, says Gina Bhawalkar, principal analyst of design and accessibility at Forrester.
“Responsible use of AI is set to transform the accessibility market and the state of digital accessibility,” she says. “We’re seeing significant advancements in using AI to improve the coverage and accuracy of automated issue detection and to generate remediation suggestions. One example is using computer vision to more accurately detect issues like text on images without sufficient contrast, or generating code suggestions for developers and suggesting 508-compliant color combinations for designers.”
“AI and machine learning techniques offer great promise for addressing digital accessibility issues,” says Ash Johnson, senior policy analyst at the Information Technology and Innovation Foundation. “There are plenty of AI tools today that can help benchmark websites and other work areas. The potential of AI is enormous in helping to make a lot of these technologies easier to use for people with disabilities.”
Currently, the NSA’s blind and low-vision community can take advantage of the latest JAWS (Job Access With Speech) assistive technology screen reader. Version 2024 includes an AI feature called Picture Smart, which describes photos to users — especially useful where the text alternative on the web page may be missing or incorrect. In the near future, the agency plans to roll out an automated testing tool that will increase accessibility testing efficiency with human-centered AI, Kron says.