This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(32, 52%, 13%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ENK%3C/text%3E%3C/svg%3E)
Definition of severity level in the link below does not match the UI. Which one is right?
I believe the UI is right.. this setting was recently modified by Azure Foundry but I need Official confirmation that the documentation is outdated. We just spend 4 hours with Azure support as our content is being blocked for no reason - they are recommend to set filters to high because they believe that will allow more severe content.
Everyone is confused and even Azure support is not clear on what is the right setting.
If set to high the allow severe content as per https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/content-filter-severity-levels?tabs=definitions
but as per UI it must be set to low to allow severe content
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(220.8, 32%, 31%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3ERS%3C/text%3E%3C/svg%3E)
Hello Nipun Kumar
Microsoft Learn confirms that the severity sliders are blocking thresholds—the level you choose is the minimum severity that will be blocked (equal or higher). Concretely, High blocks only High (allowing Low and Medium), Medium blocks Medium and High, and Low blocks Low, Medium, and High (strictest). The default is Medium for both Prompts and Completions across the four harm categories (Hate, Sexual, Violence, Self‑harm), and Safe is never filtered. This matches the portal behavior and the documentation’s wording about “equal or higher severity” thresholds: see Content filtering overview and How to configure content filters (link, link), and the Configurability table that spells out exactly what each setting blocks (link). For definitions of what counts as Low/Medium/High within each category, use the Severity level definitions page (link).
Also note that other guardrails can cause blocking independent of the severity sliders: Prompt Shields (direct jailbreak) and Protected Material (text & code) are On by default, whereas Indirect Prompt Shields is Off by default (these defaults and behaviors are described in the “Configure content filters” guide: link). If you need to permit even High severity content, that requires Modified Content Filters approval to use No filters or Annotate only modes (see the approval notes in the Configurability doc: link). In short, if your goal is to allow more severe content while still blocking the most extreme cases, set both Prompt and Completion thresholds to High; if the UI text seems to imply the opposite, treat it as ambiguous wording the effective behavior is as documented in the links above.
Hope it helps!
Thank you
Thanks @Ravada Shivaprasad for the quick response. Ok I will set everything to high again. I'll treat UI information as incorrect. Here is the screen shot of what the message says for all settings which is opposite of the documentation.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(92.8, 28.999999999999996%, 18%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EMM%3C/text%3E%3C/svg%3E)
You’re correct—the UI reflects the current behavior. Azure recently updated the content filtering logic, but the documentation hasn’t caught up yet. To allow severe content, you should set the filter to Low in the UI.
High in the UI now actually enforces stricter filtering, blocking more severe content. Azure support confirming “High allows more” is outdated advice. For now, rely on the UI settings rather than the docs, and monitor updates to the official docs for clarification.
I can confirm that I set the content filter back to high again as per the comment above and filter setting started blocking more content. I have set it back to low to allow more severe content. UI is the right setting and documentation is outdated.
Hello Nipun Kumar
Thank you for sharing the screenshot and confirming your testing results.
Based on your findings, the current Azure AI Foundry UI behavior is that setting the filter to High blocks more content, while Low allows more severe content, which is the opposite of what the Microsoft Learn documentation currently states. This indicates that the documentation is likely outdated and does not reflect the latest implementation in the portal. Your validation and screenshot provide strong evidence that the UI logic is correct and the published guidance needs to be updated.
Your approach of setting the filter to Low to allow more severe content is correct given the current UI behavior and your requirement. I will escalate this discrepancy to Microsoft engineering for official clarification and request a documentation update to avoid confusion for other users. Until we receive confirmation from the product team, please continue using the setting that meets your needs, and I will share an authoritative response as soon as it is available. Thank you for helping identify this gap it will ensure both the UI and documentation are aligned for everyone.
Hope it helps!
Thank you