Google has changed the rules it uses to guide artificial intelligence and other advanced technologies. The update removes explicit commitments that previously barred several sensitive categories of work, including weapons, some surveillance systems, and technologies that undermine human rights.
The shift replaces a list of prohibited uses with broader language about oversight, due diligence, feedback, and reducing harm. For a company deeply involved in AI research, cloud services, and generative AI products, that change alters how its public commitments are framed.
What Google Removed
Google disclosed the changes through a note added to the top of a 2018 blog post about its original AI guidelines. The note says, “We’ve made updates to our AI Principles. Visit AI.Google for the latest,” according to the source article.
The earlier principles included language saying Google would not pursue technologies that cause or are likely to cause overall harm. They also ruled out weapons or other technologies whose principal purpose or implementation is to cause or directly facilitate injury to people.
The removed language also covered technologies that gather or use information for surveillance violating internationally accepted norms. Another removed commitment applied to technologies whose purpose conflicts with widely accepted principles of international law and human rights.
The new webpage no longer presents a list of banned uses for Google AI initiatives. Instead, it says Google will use “appropriate human oversight, due diligence, and feedback mechanisms” and will work to “mitigate unintended or harmful outcomes.”
Why The Principles Matter
Google first published the principles in 2018 after internal protests over the company’s work on a US military drone program. In response, Google declined to renew the government contract and announced rules for future uses of advanced technologies such as artificial intelligence.
Those principles became a public marker for how Google said it would evaluate controversial uses of AI. They stated that Google would not develop weapons, certain surveillance systems, or technologies that undermine human rights.
In the new announcement, James Manyika, Google senior vice president for research, technology, and society, and Demis Hassabis, CEO of Google DeepMind, described the broader context for the overhaul. They pointed to the more widespread use of AI, evolving standards, and geopolitical competition over AI.
The two executives wrote that democracies should lead in AI development, guided by values including freedom, equality, and respect for human rights. They also argued that companies, governments, and organizations that share those values should work together on AI that protects people, supports growth, and supports national security.
Employee Concerns And Internal Review
Multiple Google employees told WIRED they were concerned about the changes. Parul Koul, a Google software engineer and president of the Alphabet Union Workers-CWA, said it was concerning to see Google drop its ethical AI commitment “without input from its employees or the broader public.”
After the 2018 principles were introduced, Google created two teams to review whether company projects aligned with those commitments. One team focused on core operations, including search, ads, Assistant, and Maps. Another focused on Google Cloud offerings and customer deals.
The unit focused on Google’s consumer business was split up early last year while the company pushed to develop chatbots and other generative AI tools to compete with OpenAI.
Timnit Gebru, a former colead of Google’s ethical AI research team who was later fired from that position, questioned whether Google had lived up to the principles. Three former Google employees who reviewed projects under the principles said the work could be difficult because the principles were open to different interpretations and because leaders pushed business priorities.
What Still Remains In Policy
Google still has harm-related language in its official Cloud Platform Acceptable Use Policy, which covers various AI-driven products. That policy forbids violating the legal rights of others and engaging in or promoting illegal activity, including terrorism or violence that can cause death, serious harm, or injury to individuals or groups.
Google Cloud’s Terms of Service also prohibit applications that violate the law or “lead to death or serious physical harm to an individual.” Rules for some consumer-focused AI services also ban illegal uses and some harmful or offensive uses.
Project Nimbus remains part of the debate around how those policies apply. When asked how the Cloud Platform Acceptable Use Policy fits with the cloud computing contract with the Israeli government, which has benefited the country’s military, Google has said the agreement “is not directed at highly sensitive, classified, or military workloads relevant to weapons or intelligence services.”
Google spokesperson Anna Kowalczyk told WIRED in July that the Nimbus contract covers workloads running on commercial cloud by Israeli government ministries, which agree to follow Google’s Terms of Service and Acceptable Use Policy.
A Shift From Bans To Governance
The practical change is not just in wording. Google’s earlier AI principles named categories the company said it would avoid. The revised version gives the company more room to consider sensitive uses under broader governance commitments.
Google now lists its goals as bold, responsible, and collaborative AI initiatives. Phrases such as “be socially beneficial” and “scientific excellence” are gone, while a reference to respecting intellectual property rights has been added.
That makes the new Google AI principles less like a public boundary line and more like a framework for review. The company says it will stay consistent with widely accepted principles of international law and human rights, but the clearest change is that weapons and surveillance are no longer separately listed as banned categories.